Reflection and the Stability of Belief : Essays on Descartes, Hume, and Reid 9780199709311, 9780195368765

A unifying theme of Loeb's work is epistemological - that Descartes and Hume advance theories of knowledge that rel

251 113 2MB

English Pages 388 Year 2010

Report DMCA / Copyright

DOWNLOAD PDF FILE

Recommend Papers

Reflection and the Stability of Belief : Essays on Descartes, Hume, and Reid
 9780199709311, 9780195368765

  • 0 0 0
  • Like this paper and download? You can publish your own PDF file online for free in a few minutes! Sign Up
File loading please wait...
Citation preview

reflection and the stability of belief

This page intentionally left blank

REFLECTION AND THE STABILITY OF BELIEF Essays on Descartes, Hume, and Reid

Louis E. Loeb

2010

Oxford University Press, Inc., publishes works that further Oxford University’s objective of excellence in research, scholarship, and education. Oxford New York Auckland Cape Town Dar es Salaam Hong Kong Karachi Kuala Lumpur Madrid Melbourne Mexico City Nairobi New Delhi Shanghai Taipei Toronto With offices in Argentina Austria Brazil Chile Czech Republic France Greece Guatemala Hungary Italy Japan Poland Portugal Singapore South Korea Switzerland Thailand Turkey Ukraine Vietnam

Copyright © 2010 by Oxford University Press, Inc. Published by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016 www.oup.com Oxford is a registered trademark of Oxford University Press All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Library of Congress Cataloging-in-Publication Data Loeb, Louis E. Reflection and the stability of belief: essays on Descartes, Hume, and Reid / Louis E. Loeb. p. cm. Includes bibliographical references and index. ISBN 978-0-19-536876-5; 978-0-19-536875-8 (pbk.) 1. Reason. 2. Belief and doubt. 3. Descartes, René, 1596–1650. 4. Hume, David, 1711–1776. 5. Reid, Thomas, 1710–1796. I. Title. BC177.L635 2010 190—dc22 2009038194

9 8 7 6 5 4 3 2 1 Printed in the United States of America on acid-free paper

For Tully

This page intentionally left blank

acknowledgments

culling the original acknowledgments for the individual essays, it will be obvious that I have accumulated substantial intellectual debts to David Hills and David Velleman. My work could never have progressed as far as it has without their insight and commitment of time and interest. I have benefited from Frederick Schmitt’s gift for synthesizing epistemology past and present and from his selflessness in responding to inquiries. His impact on my work is apparent in the Introduction. I owe much as well to Janet Broughton. Her subtlety in interpreting Descartes and Hume has been an inspiration. Lawrence Sklar has for many years tolerated much random pestering, as I have leaned on him better to grasp basic issues in epistemology and their interaction with the history of modern philosophy. James Joyce has also been an exceptionally helpful sounding board in this regard. I enjoyed our coteaching a proseminar in 2008, and learned a lot. Edwin Curley’s first book, on Spinoza, influenced my earliest efforts in the history of modern philosophy; I have been fortunate to have him as a colleague and resource later in my career. In recent years, Victor Caston has cheerfully commented on drafts. He has also been generous with his time in providing editorial suggestions, as well as practical advice about optical character recognition software that proved indispensable in bringing this volume to fruition. Alex Silk has contributed to the volume in a variety of ways. He worked tirelessly and with great care to help standardize style, citations of primary sources, and the author-date reference system and to verify that the chapters preserve the contents of the original articles. He prepared the extensive index. In addition, his suggestions led to considerable improvement in the Introduction.

Others have been sources of heartening encouragement—Steve Darwall, Robert Ferrell, Gary Fuller, Debra Nails, and Peter Ohlin, my editor at Oxford University Press, New York. Readers for the press made encouraging and valuable suggestions that were much appreciated. I wish I had been able to follow them all. Tad Schmaltz kindly offered help with this project, even before coming on board as a colleague at Michigan. I thank Tom Beauchamp for taking time to pursue questions about publication details relating to the Clarendon and Oxford Philosophical Text (OPT) editions of Hume. The University Library’s Interlibrary Loan and FAST delivery services have been amazingly efficient, greatly reducing the demands of preparing this volume. The University of Michigan, the Office of the Vice-President for Research, and the College of Literature, Science, and the Arts provided sabbatical and fellowship leave, and the Department of Philosophy research support. This volume would not have been possible without this assistance. Over the years, members of the profession whom I had not previously met have told me at conventions and conferences how much my first book influenced their thinking. These have been among the most gratifying moments of my career. Special thanks go to my family. To this day, I use the copy of the Treatise that Sophie rebound for me many years ago. Her good cheer and wit are always a delight. Simon was good to indulge my grousing about tasks at hand. I enjoyed philosophical conversation with Gabriel during the final stages of preparing this volume. His vision of Hume working at a laptop has often given me a much needed chuckle. Tully has been unfailing in her support of this and other projects—no matter how much they have imposed on time together, or how many “progress” reports she has had to endure. I will be forever grateful.

viii

acknowledgments

contents

The Essays

xi

Abbreviations for Editions of Seventeenthto Nineteenth-Century Works

xiii

References to Hume

xvii

Introduction

3

1. Is There Radical Dissimulation in Descartes’ Meditations? (1986)

34

2. The Priority of Reason in Descartes (1990)

58

3. The Cartesian Circle (1992)

89

4. Sextus, Descartes, Hume, and Peirce: On Securing Settled Doxastic States (1998)

117

5. Integrating Hume’s Accounts of Belief and Justification (2001)

143

6. Hume’s Explanations of Meaningless Beliefs (2001)

165

7. Hume on Stability, Justification, and Unphilosophical Probability (1995)

184

8. Hume’s Agent-centered Sentimentalism (2003)

214

9. What Is Worth Preserving in the Kemp Smith Interpretation of Hume? (2009)

245

10. Psychology, Epistemology, and Skepticism in Hume’s Argument about Induction (2006)

270

11. Locke and British Empiricism (forthcoming)

288

12. The Naturalisms of Hume and Reid (2007)

308

Bibliography

333

Index

353

x

contents

the essays

quotations are retained as they appeared in the source articles, though systems of reference to seventeenth- and eighteenth-century works have been altered in light of the scheme of abbreviations explained later. The sources for quotations of Hume are explained in an early footnote in each of chapters 2 and 4 through 12. Unless otherwise noted, the chapters preserve the titles and section numbers of the original articles. With the exception of chapters 5 and 7, the numbers of notes correspond to those carrying the same content in these sources. Notes have been lightly rewritten in the interest of a uniform author-date format. I have updated references to articles or books forthcoming when a source article was published. Occasional changes or additions in notes are placed in square brackets. The chapters correct typographical and other obvious errors in the original articles (hopefully without introducing a greater number of new errors) and are amended for consistency in style and conventions throughout the volume. I am pleased to acknowledge permissions for the articles to appear in this volume. (Full bibliographical references, where not included here, may be found in the bibliography.) Chapter 1. Louis E. Loeb, “Is There Radical Dissimulation in Descartes’ Meditations?” Copyright © 1986, The Regents of the University of California, published by the University of California Press. Chapter 2. “The Priority of Reason in Descartes,” in The Philosophical Review, vol. 99, no. 11, pp. 3–43. Copyright © 1990, Sage School of Philosophy, Cornell University. All rights reserved. Used with permission of the publisher, Duke University Press.

xi

Chapter 3. “The Cartesian Circle,” reprinted with permission of Cambridge University Press. Chapter 4. “Sextus, Descartes, Hume, and Peirce,” reprinted with permission of Noûs and Blackwell Publishing. Chapter 5. “Integrating Hume’s Accounts of Belief and Justification,” reprinted with permission of Philosophy and Phenomenological Research and Blackwell Publishing. Chapter 6. “Hume’s Explanations of Meaningless Beliefs,” reprinted with permission of Philosophical Quarterly and Blackwell Publishing. Chapter 7. “Hume on Stability, Justification, and Unphilosophical Probability,” reprinted with permission of Journal of the History of Philosophy and Johns Hopkins University Press. Chapter 8. “Hume’s Agent-centered Sentimentalism,” from Philosophical Topics, vol. 31, nos. 1 & 2 (Fayetteville, University of Arkansas Press, 2003). Used with permission of the publisher, www.uapress.com. Chapter 9. “What Is Worth Preserving in the Kemp Smith Interpretation of Hume?” British Journal for the History of Philosophy (http:// www.informaworld.com), September, 2009 (vol. 17, no. 4), 769–97. Reprinted with permission of Taylor & Francis. Chapter 10. “Psychology, Epistemology, and Skepticism in Hume’s Argument about Induction,” reprinted with minor alterations by permission of Synthese and Springer. Chapter 11. “Locke and British Empiricism,” forthcoming in Matthew Stuart, ed., A Companion to Locke, appears here by permission of Blackwell Publishing. Chapter 12. “The Naturalisms of Hume and Reid,” reprinted with permission of the American Philosophical Association.

xii

the essays

abbreviations for editions of seventeenth- to nineteenthcentury works

where there is no risk of loss of clarity, the abbreviation for an edition or work is omitted after its initial inclusion within a string of references to the same source in a single paragraph. Abs.

AT

B-00 B-98 BNS

Br

An Abstract of . . . A Treatise of Human Nature. Page references are to the Selby-Bigge and Nidditch edition at T below; quotations are from the source cited in the individual chapters. Charles Adam and Paul Tannery, eds., Oeuvres de Descartes, vols. 1–11 (Paris: Libraire Philosophique J. Vrin, 1964). Where an English source is listed after an AT reference, I have departed from the cited translation. References to the specific paragraphs of the Meditations follow the paragraph divisions of the second Latin edition, 1642, as edited by Adam (AT 7); this is the edition translated in CSM and HR. Thomas L. Beauchamp, ed., An Enquiry concerning Human Understanding (Oxford: Clarendon Press, 2000). Thomas L. Beauchamp, ed., An Enquiry concerning the Principles of Morals (Oxford: Clarendon Press, 1998). T. L. Beauchamp, D. F. Norton, and M. A. Stewart, Humetext 1.0 (Washington, D.C.: Georgetown University, 1990). The text in this electronic edition of Hume’s works represents an early stage in the editorial process culminating in the Clarendon edition of Hume. Craig B. Brush, ed. and trans., The Selected Works of Pierre Gassendi (New York: Johnson Reprint, 1972).

xiii

C

CSM

CSMK

CTR CW

DHP

DNR EHU

EIP

EL

EPM

Essay Essays HR

IHM

xiv

John Cottingham, trans., Descartes’ Conversation with Burman (Oxford: Clarendon Press, 1976). References are to Cottingham’s numerical divisions of the text. John Cottingham, Robert Stoothoff, and Dugald Murdoch, trans., The Philosophical Writings of Descartes, vols. I and II (Cambridge: Cambridge University Press, 1984). John Cottingham, Robert Stoothoff, Dugald Murdoch, and Anthony Kenny, trans., The Philosophical Writings of Descartes, vol. III, The Correspondence (Cambridge: Cambridge University Press, 1991). Paul Wood, ed., The Correspondence of Thomas Reid (University Park: Pennsylvania State University Press, 2002). John M. Robson, ed., Collected Works of John Stuart Mill (Toronto: University of Toronto Press), vol. 9, An Examination of Sir William Hamilton’s Philosophy (1979) and vol. 11, Essays on Philosophy and the Classics (1978). Three Dialogues between Hylas and Philonous, in T. E. Jessop, ed., The Works of George Berkeley, vol. 2 (London: Thomas Nelson, 1949). Cited by dialogue and page number. Norman Kemp Smith, ed., Dialogues concerning Natural Religion (Oxford: Clarendon, 1935). Cited by part and page number. An Enquiry concerning Human Understanding, in L. A. Selby-Bigge, ed., Enquiries concerning the Human Understanding and concerning the Principles of Morals, third edition, by P. H. Nidditch (Oxford: Clarendon Press, 1975). Page references are to this edition; quotations are from the source cited in the individual chapters. Derek R. Brookes, ed., Essays on the Intellectual Powers of Man (University Park: Pennsylvania State University Press, 2002). Proper names converted from all upper case. Ferdinand Tönnies, ed., The Elements of Law, Natural and Politic, second edition, with a new introduction by M. M. Goldsmith (London: Frank Cass, 1969). Cited by part, chapter, and section. An Enquiry concerning the Principles of Morals, in Selby-Bigge and Nidditch at EHU above. Page references are to this edition; quotations are from the source cited in the individual chapters. P. H. Nidditch, ed., An Essay concerning Human Understanding (Oxford: Clarendon Press, 1975). Eugene F. Miller, ed., Essays, Moral, Political, and Literary (Indianapolis: Liberty Classics, 1987). Elizabeth S. Haldane and G. R. T. Ross, trans., The Philosophical Works of Descartes, vols. 1–2 (Cambridge: Cambridge University Press, 1970). Derek R. Broackes, ed., An Inquiry into the Human Mind on the Principles of Common Sense (University Park: Pennsylvania State University Press, 1997).

abbreviations

IMG

K LDH Lev.

LO

LRF

NN O PEP

PHK PN PRLS

T

Works

Francis Hutcheson, An Inquiry Concerning the Original of Our Idea of Virtue or Moral Good, in An Inquiry into the Original of Our Ideas of Beauty and Virtue, in Two Treatises (Dublin: Will. and John Smith, 1725). Anthony Kenny, trans. and ed., Descartes, Philosophical Letters (Oxford: Oxford University Press, 1970). J. Y. T. Grieg, ed., Letters of David Hume, 2 vols. (Oxford: Clarendon Press, 1923). Edwin Curley, ed., with an introduction, Leviathan, with Selected Variants from the Latin Edition of 1668 (Indianapolis: Hackett, 1994). Cited by chapter and paragraph. Paul J. Olscamp and Thomas M. Lennon, eds., The Search after Truth, Elucidations of The Search after Truth (Cambridge: Cambridge University Press, 1997). The Search is cited by book, part, chapter, and section, followed by LO page number. Alexander Broadie, ed., Thomas Reid on Logic, Rhetoric and the Arts; Papers on the Culture of the Mind (University Park: Pennsylvania State University Press, 2005). David Fate Norton and Mary J. Norton, eds., A Treatise of Human Nature, vol. 1 (Oxford: Clarendon Press, 2007). Paul J. Olscamp, trans., Discourse on Method, Optics, Geometry, and Meteorology (Indianapolis: Bobbs-Merrill, 1965). James Mackintosh, Dissertation on the Progress of Ethical Philosophy, second edition (Edinburgh: Adam and Charles Black, 1837). A Treatise Concerning the Principles of Human Knowledge, in Jessop at DHP. Cited by section number in Part I. Philosophical Notebooks, in Jessop at DHP, vol. 1. Cited by entry number. Paul Wood, ed., Thomas Reid on the Animate Creation; Papers Relating to the Life Sciences (University Park: Pennsylvania State University Press, 2002). L. A. Selby-Bigge, ed., A Treatise of Human Nature, second edition by P. H. Nidditch (Oxford: Clarendon, 1978). Page references are to this edition; quotations are from the source cited in the individual chapters. The Works of John Locke, ten vols. (London: Thomas Tegg, 1823).

abbreviations

xv

This page intentionally left blank

references to hume

in 2006, oxford university Press began to incorporate Selby-Bigge and Nidditch edition—see EHU and T—page numbers into the margins of the Clarendon and Oxford Philosophical Text (OPT) editions and printings, cloth and paper, of An Enquiry concerning the Human Understanding and An Enquiry concerning the Principles of Morals (edited by Tom L. Beauchamp) and of A Treatise of Human Nature (edited by David F. Norton and Mary J. Norton). This generally nullifies the need to provide cumbersome “dual references”—to Selby-Bigge page numbers and to Clarendon and OPT paragraph numbers—for Hume citations. I do provide dual references for the passages in the Appendix to the Treatise (at Selby-Bigge and Nidditch, pages 627–33) that Norton and Norton insert into Book I proper. Two passages of importance to my interpretation do not appear in the text of the Treatise in the Clarendon and OPT editions: the final paragraph at Selby-Bigge and Nidditch 123 and Selby-Bigge and Nidditch 371n. For an explanation of the Clarendon treatment of 123, see Norton and Norton 2007 2.650 (emendation at 84.41–85.44) and 2.751 (annotation to 84.41– 85.44 of the Appendix); for 371n., see Norton and Norton 2000, pages 458 (annotation to 1.3.9.19) and 515 (final annotation to 2.2.7.6), or Norton and Norton 2007 2.471, 594, and 653 (emendation at 240.4).

xvii

This page intentionally left blank

reflection and the stability of belief

This page intentionally left blank

Introduction

this volume brings together a dozen of my articles in the history of modern philosophy published since 1986. The focus is the contribution of psychological claims to epistemology in Descartes, Hume, and Reid and the decline in the fortune of reason and reflection as gauged by the later figures. Both Descartes and Hume advance theories of knowledge that adopt one or another member of a cluster of psychological properties of beliefs (unshakability, stability) as the goal of inquiry and the standard for assessing belief-forming mechanisms. There is thus a surprising affinity in the constructive epistemological theories of two archetypal “rationalist” and “empiricist” figures. Chapter 4, “Sextus, Descartes, Hume, and Peirce: On Securing Settled Doxastic States” (“SDHP”), which best gives the flavor of the unity in my approach to diverse figures, might be read first. I have by and large arranged the papers by date of publication to reflect the development of my views. A few are placed out of sequence to bring out topical interconnections. Because I have worked successively on epistemology in Descartes, Hume, and Reid, the progression of chapters correlates well with the historical chronology. The papers fall into overlapping groups on three themes: the role of unshakability in Descartes (chapters 1–4), the stability interpretation of Hume and its application to some interpretive problems (chapters 4–9), and a broader class of interpretations of Hume in terms of epistemologies that downplay the role of reflection (chapters 9–12). Two chapters within this third group consider Hume in relationship to predecessors and successors, most especially Locke and Reid.

3

The last five essays postdate my 2002 book, Stability and Justification in Hume’s Treatise (SJHT).1 My lengthy, 2004 defense of SJHT is not included, though this introduction provides an orientation to the collection’s contents, taking into account all my recent work. I do not attempt commentary on the abundant, excellent literature since the first publication of many of the papers.

1. Two Epistemological Projects (Chapter 10) Descartes and Hume pursued dramatically different epistemological projects. I do not have in mind their assigning pride of place to reason and senseperception, respectively. I am thinking instead of a more fundamental difference in their conceptions of the principal subject matter of a theory of the understanding. I first develop these themes in papers published in 2006, “Psychology, Epistemology, and Skepticism in Hume’s Argument about Induction” (“Psychology”), and 2008, “Inductive Inference in Hume’s Philosophy.”2 Descartes seeks to explain how knowledge is possible for a suitably idealized cognizer, one who is fully reflective, consciously articulates the complete range of considerations relevant to a judgment, draws out all their implications, and so forth. Descartes’ quarry is the best possible reflective knowledge, scientia or “scientific knowledge.” By contrast, Hume seeks to explain how full-fledged knowledge is possible for quite ordinary cognizers— natural philosophers, certainly, but also relatively unreflective adults, children, and nonhuman animals. Descartes’ bifurcation of the human and animal realms, with animals lacking reason, minds, and even consciousness, is no obstacle to his study of reflective knowledge.3 Hume highlights the reason of animals, the topic of the concluding section of Treatise I.iii and of one of twelve divisions comprising the first Enquiry (“the Enquiry”). This material is essential to his epistemological enterprise. Hume’s “system concerning the nature of the understanding” should be put to “this decisive trial”—“whether it will equally account for the reasonings of beasts as for those of the human species” (T 177; cf. EHU 104).4 An animal (T 178), just as a human (103),

1. I have omitted papers in pure metaphysics (chiefly, my 1992b and 2005) and a number of articles (1991, 1995b, and 1997) substantially reproduced in my 2002 book. (The bibliography for this volume lists all my papers and reviews in the history of modern philosophy.) The middle group of chapters does include articles (chapters 5–7) either central to the interpretation of SJHT or important to conveying its fruitfulness. Chapters 6 and 7 of SJHT, on the role of constancy and coherence in I.iv.2 and on ways in which Hume’s psychology of belief formation is underspecified, are not represented in the present volume. 2. This volume includes only the 2006 paper; composed more recently, it better reflects my current views on Hume on induction. 3. For a critical survey of the view that Descartes allows animal consciousness, see Hatfield 2008, esp. 420–22. 4. In this chapter, quotations of the Treatise and the Enquiry are based on the volumes edited by Norton and Norton (NN) and Beauchamp (B-00) in the Clarendon edition of Hume’s works.

4

reflection and the stability of belief

“foresees” future events on the basis of causal inference. Habit or custom thus “supplies the place of . . . reflection” (93). Animals and more reflective creatures have equal standing, qua the character of the knowledge they possess. In “The Naturalisms of Hume and Reid” (“Hume and Reid”), I argue that Hume is aligned with Reid in this respect. A recurrent theme in Hume, and Reid, is that children and most adults are ignorant of any cogent arguments that might exist for fundamental beliefs, such as belief in the external world. What strategies, short of declaring reflective and unreflective knowledge different in kind, are then available within the Cartesian project for explaining knowledge of less reflective creatures?5 Perhaps ordinary folk have knowledge in virtue of the availability of fully reflective knowledge, knowledge they “could” possess were their faculties better developed and employed. Even if this strategy can account for the knowledge of normal children, whose powers of reflection mature, reflective knowledge is scarcely “available” to animals. Or perhaps ordinary creatures possess knowledge in virtue of holding a belief system that constitutes a suitable approximation to scientia. The challenge is to spell out the requirements for a “sufficient approximation” in a way that the ideal does real work, without undercutting the extensiveness of everyday knowledge. These difficulties aside, Hume would object that both strategies are misguided: they attempt an indirect explanation of mundane knowledge, as derivative from the knowledge of ideal cognizers. Hume’s program—to explain how unreflective beings possess knowledge in their own right—has a downgrading of the role of reflection for a consequence. Hume also finds Cartesian explanations of knowledge hopeless, mistakenly presupposing that the Cartesian ideal is in principle achievable. There are no necessary truths to authorize inductive inference, Hume’s central case. Descartes could offer an argument within the constraints of his foundationalism, possibly through Divine immutability, but not a good argument. This does not imply that one cannot attain inductive knowledge, only that such knowledge cannot be based on a cogent demonstrative or non-question-begging argument. Since Cartesian knowledge of the uniformity of nature is beyond the reach of even fully reflective inquirers, the inductive knowledge of ordinary cognizers cannot derive from that available to idealized epistemic agents. This, not that all inductive belief is unjustified, is the lesson of I.iii.6. Distinguishing the projects of explaining reflective versus routine knowledge helps illuminate why (as I argue) Descartes focuses on unshakability, Hume on a less stringent requirement of stable belief. A belief could be stable, though susceptible to being overturned by considerations not yet in play. Unshakability is perhaps a fitting objective for a Cartesian inquirer who can recognize and process all relevant evidence, but not for individuals with limited processing capacities and inputs to date. Stability, however, is within their reach. 5. For the suggestion that unreflective knowledge is of an inferior kind, see Sosa 1997, 422, 426–27. Cohen also discusses this way of proceeding (2002, 326–27).

introduction

5

Hume’s and Descartes’ differing views of the power of reason ramify further. Though Descartes addresses the role of authority and faith, an inventory of cognitive faculties—intellection or reason, sense-perception, and memory, together with the imagination, an imagistic by-product of sense-perception— is preset. The problem is not to identify the faculties that secure knowledge, but to explain their interdependencies and to show that they are sufficient to combat skepticism. Hume sees reason as operating in a highly restricted domain. For him, there is no successful demonstrative argument—whether Cartesian through the Divine attributes, or Lockean by way of necessary connections between real essences—that justifies beliefs about the unobserved. Reason cannot supplement sense-perception and memory to account for knowledge of unobserved matters of fact. In the Treatise, habit or custom is the primary locus of an additional cognitive faculty, producing knowledge of the unobserved. Its products are rooted in the imagination, a faculty of association that preserves (in the case of memory) and produces belief. All beliefs due to the imagination, broadly conceived in this way, can look “trivial,” if compared to those generated by a nonassociative faculty of reason that grasps necessary connections (cf. T 265). Yet, a general faculty of association is our only hope for knowledge of the unobserved. In this vein, only some imaginative “principles” are “trivial”—those that “take place in weak minds” and “may easily be subverted” (224–25). These include “whimsies and prejudices” and constitute “the imagination” in a narrow sense (117n.). Association by the relation of cause and effect is “general and more establish’d” (267). Together with demonstration, it constitutes “reason” (117–18n.) or “the understanding” (267, 371n.).6 This faculty, consisting of propensities that are “permanent, irresistible, and universal” (224–25), is far from trivial, psychologically. Inductive (and memory) beliefs pull their own weight—at their strongest, they equal or approximate the irresistibility of Cartesian clear and distinct perception (reason and introspection). Hume’s demarcation among nondemonstrative belief-forming mechanisms has the effect of expanding the Cartesian (and Lockean) inventory of cognitive faculties.

2. Obstacles to Interpretation It has not been easy for commentators to discern Hume’s project of explaining unreflective knowledge. It can seem questionable whether he is engaged in a normative epistemological enterprise, much less offering anything approximating a theory. The structure of Book I of the Treatise—the inclusion of Part ii, on space and time, and the character and sequence of topics in Part iv—is less than perspicuous. Hume scatters epistemological remarks in the course of his exposition of associationist or psychological claims.

6. For a textual note on Treatise 371n., see “References to Hume,” this volume.

6

reflection and the stability of belief

Locating the heart of his epistemological position requires threading together bits in I.iii.9–10, 12–13, and 15–16, the first paragraph of I.iv.4, and II.iii.9–10. His main interest is the science of human nature. Further, Book I is rich in dazzling philosophical developments—the examination of inductive inference (I.iii.6), the occurrentist account of belief (I.iii.7), the constant conjunction analysis of causation and the critique of necessary connection (I.iii.14), the exploration of belief in the external world (I.iv.2), and the bundle theory of the self and the causal theory of personal identity (I.iv.6). Books and courses canvassing this hit parade, even with such backfilling as might seem required (for example, the Lockean stagesetting in I.i), often skirt the sections indicative of Hume’s positive epistemology. In addition, readers disposed to a skeptical reading of I.iii.6 find ample material in Book IV to reinforce assaying Hume a skeptic. Taking I.iv.1 (“Of scepticism with regard to reason”), I.iv.4, and I.iv.7 at face value leaves little space for an interpretation on which Hume advances an epistemology that explains knowledge, rather than destroys it. As it happens, these sections run out of control (SJHT, §§VII.2–3). Had Hume been a doctoral student in our time, his advisors would have pressed him to revise, on the ground that there must be an outright mistake in the I.iv.1 argument for the reduction of probability to zero, and in the Berkelian argument for the inconceivability and impossibility of matter in I.iv.4. A fondness for provocation is already on display in the opening section of the Treatise: Hume lays out the “copy principle” and the missing shade, announcing the latter “scarce worth our observing.” We can be grateful supervisors did not press Hume to assess the counterexample’s scope and to reconsider arguments in Part iv. His masterpiece might not have seen the light of day, or it might have been beyond recognition. It remains for interpreters to come to terms with Hume’s mischievousness. Descartes’ intellectual personality was dramatically different. Epistemology is front and center. The sequence of topics in the Meditations and Principles mirrors the foundational order, as Descartes conceives it. Foundationalism suited his structured and deliberate style. Descartes was as guarded, methodical, and sober as Hume was freewheeling. It is easy to picture Hume revising for style and dramatic effect, Descartes for substance and for tactical advantage. Prudence shapes Hume’s Dialogues concerning Natural Religion and the Natural History of Religion, but its impact on these specialized works is comparatively transparent. As sketched in §4 of chapter 1, Descartes’ solicitousness of the church pervades the matter and publication history of main works—The World, the Meditations, and the Principles of Philosophy. Because of his theological caution, Descartes presents interpretive difficulties of his own. We might wonder why the causal or anthropological arguments for God are placed first, with the ontological argument held in reserve.7 There are more far-reaching questions. Descartes unabashedly 7. For discussion and one recent explanation, see Curley 1978, 156–64, and Menn 1998, esp. ch. 6, respectively.

introduction

7

grounds human knowledge in Divine veracity and the laws of motion in Divine immutability. Yet, God’s free creation of the eternal truths does not emerge in the Discourse on the Method. In the Meditations, it arguably makes no clear appearance and plays no obvious role, though it surfaces in the Fifth and Sixth Replies. The situation with the Principles is similar. There is a suggestion of the creation doctrine in a list of God’s properties at I.22— “eternal, omniscient, omnipotent, the source of all goodness and truth, the creator of all things, and . . . infinite” (CSM 1.200: AT 8A.13). Descartes does not draw out any metaphysical implications, not even at I.48–50 and 75, where we find the eternal truths terminology. All told, some theses that rely on the Divine attributes are on prominent display, but the view that putatively necessary truths are dependent on God’s will is far from the limelight. Attempting to understand how (or whether) this doctrine interacts with the deceiver hypothesis and with the claims that God, of necessity, exists and is veracious and immutable, leads to mind-boggling interpretive and philosophical puzzles.8 Descartes offers little assistance; he does not apply himself to these questions in works published in his lifetime.

3. Arguments, Motivations, and Insincerity My first book, From Descartes to Hume: Continental Metaphysics and the Development of Modern Philosophy (1981), gave special attention to causation and substance. I often found the gap between the quality of a philosopher’s argument and the strength of his commitment to its conclusion too great to warrant supposing that the proffered line of reasoning explains the figure’s holding the conclusion. In such situations, historians bring to bear a principle of charity, coupled with a contextual or analytic impulse. It is suggested that an argument fares much better, if only we supply either unstated assumptions that were common property at the time or suppressed premises— ones consistent with a figure’s fundamental principles—that were “at the back of his mind.” I became convinced that these reconstructive techniques, though often essential tools, do little to illuminate how Descartes could hold that animals are not conscious and lack minds, or why Malebranche embraced occasionalism, or Berkeley a variant of that view. Our understanding of these figures 8. For commentators who read the doctrine of the eternal truths into the dialectic of the deceiver or the circle, see Bréhier 1937; Etchemendy 1981; Gaukroger 1995, 316–18, 340; Mawson 2001; and Della Rocca 2005. Also see Murdoch 1999. Bennett 1977 sees the doctrine as largely immaterial to these issues; Koyré 1922, 19–21, and Nadler 1987 think it destructive of much in Descartes’ system. For views on the doctrine’s implications for the ontological argument, see Curley 1984, §7; Bennett 1977, §5, and 2001, 2.66–69; Pessin 2006; and Schmaltz 2008, 187–90. For the relationship between the eternal truths and the laws of motion, see Broughton 1987; Dutton 1996; Pavelich 1997; Della Rocca 1999, esp. 62–70; Schmaltz 2003, esp. §§1–2; and Kaufman 2005, esp. §4. Hatfield 1993 and Menn 1998, ch. 8, offer sustained hypotheses about the role of the doctrine in Descartes’ physics.

8

reflection and the stability of belief

is best enhanced by acknowledging that these claims were held to buttress a theocentric worldview or specific religious doctrine (that of human immortality, in the case of Descartes’ denying animal minds or souls). We should narrow the gap between argument and conclusion where we can but not pretend to close it. In subsequent work on Descartes’ epistemology, I found stronger claims attractive: not only that motives help explain conclusions where arguments fall short but also that Descartes was insincere in offering his proofs of the existence of God and in appealing to Divine veracity to validate clear and distinct perception. In a 1988 paper, “Was Descartes Sincere in His Appeal to the Light of Nature?” (not included in the present volume), I argued that Meditations III and IV suggest dissimulation. First, Descartes’ argument seems plainly to beg the question against the hypothesis of a powerful being who deceives us with respect to our most certain beliefs—even beliefs in elementary arithmetic. How could Descartes have been unaware of this defect in his argumentative strategy? Second, the arguments for the existence of God depend on an indefensible causal principle—that there must be at least as much formal perfection in the efficient and total cause of an idea as objective perfection in the idea itself—which cannot be explained as a medieval inheritance. How could Descartes have been unaware of the glaring substantive deficiencies in the argument? Third, Descartes appeals to the “natural light” in declaring dubious metaphysical theses, such as the causal principle, not only true but also evident or manifest. How could Descartes have thought the causal principle for ideas certain, more so than simple arithmetic, courtesy of the natural light? The dissimulation hypothesis offers a unified set of answers to these questions. In fashioning arguments for the existence of God he does not endorse, Descartes invokes the natural light to introduce a premise gerrymandered from Scholastic materials. More generally, the light of nature makes its debut as a source of metaphysical principles at precisely those junctures where appeal to clear and distinct perception would render the circularity blatant. Calling upon the natural light is a device intended to obscure the argument’s procedural and substantive defects. So I argued, hoping to show that an interpretation along these lines merits exploration. Dissimulation theories, whether in regard to Descartes’ epistemology or his mind-body metaphysics, have gained little traction. Historians presume that attributions of insincerity indicate defects of character in our philosophical forerunners. Even more than auxiliary appeals to motives, dissimulation hypotheses also threaten to undermine the reconstructive enterprise: it seems pointless to search out historical or philosophical sources for positions a figure feigned to adopt, leaving room only for historical and biographical investigation of the pressures toward insincerity. This reaction is an oversimplification. Beyond identifying the forces propelling the insincerity, a dissimulation hypothesis worthy of attention must provide a rationale for delimiting the reach of the pretense in a figure’s system. I see my 1986 paper, “Is There Radical Dissimulation in Descartes’ Meditations?”

introduction

9

(“Radical”), as an illustration of the type of work required to discharge this obligation. In the case of my hypothesis of dissembling in the two middle Meditations, the problem is severe: what remains of Descartes’ theory of knowledge, absent the Divine validation of the cognitive faculties? “Radical” puts forward an answer, one that I developed in a series of papers in the 1990s, “The Priority of Reason in Descartes” (“Priority”), “The Cartesian Circle” (“Circle”), and “SDHP.” These are the first four chapters of this collection. This body of work progressed in a way that led to a promising line of solution to the problem of the circle (§5), undercutting the first of the three considerations that led me to a dissimulation theory. I now take the availability of this solution to weigh heavily against dissimulation, though the later articles do not help explain why Descartes invoked the light of nature or how he could have been on board with the causal principle for ideas. These puzzles remain. In any event, in pursuing dissimulation I came to identify in Descartes’ epistemology a component—a naturalistically conceived hierarchy of cognitive faculties—that does not depend on a proof of the nondeceiving God.

4. The Hierarchy of Cognitive Faculties (Chapters 1–2) If Descartes was not sincere in his appeal to Divine veracity, what is his response to skepticism? In “Radical,” following Harry Frankfurt (1970), I take the deceiver hypothesis in Meditation I to be confined to beliefs based on the senses alone—belief in the material world, but also belief in mathematics as the empiricist conceives it, as a body of generalizations based on experience. In Meditation III, the deceiver hypothesis extends to beliefs based on reason, or clear and distinct perception. I argue that disanalogies in Descartes’ formulations of the two hypotheses indicate that the latter is but a pretext for locating an epistemological role for God. In treating Descartes’ response to skepticism, a dissimulation theory can then divide and conquer. I locate in Descartes a nontheological response to skepticism about senseperception. Descartes maintains that whereas beliefs based on reason cannot be corrected by any other faculty, beliefs based on sense-perception are susceptible to correction by reason. This commitment, surfacing early in Meditation III, is independent of the proof of Divine veracity. A belief is maximally reasonable just in case it has survived all possible tests for correction. Some perceptual beliefs—that bodies possess properties exactly resembling sensory experiences of secondary qualities, and that there are vacuums—do not pass muster. The point of the skepticism in Meditation I is that any belief based on sense-perception is susceptible to correction; the heart of Descartes’ response is the claim that belief in the existence of the material world withstands all possible tests for correction. There is no room for a parallel worry about reason, since it does not admit of correction.

10

reflection and the stability of belief

In “Priority,” I take up a pressing question: on what basis does Descartes assert a hierarchical ordering of reason and sense-perception? “Radical,” without addressing this question, assumes that “correction” is a matter of showing a belief false. This invites the thought that Descartes grounds an asymmetrical relationship between reason and sense-perception by way of the truth rule: whatever one clearly and distinctly perceives is true, so that reason has no need of correction. Since Divine veracity grounds the truth rule, the hierarchical conception of the faculties is dependent on the proof of the nondeceiving God after all. This tempting explanation of the hierarchy does not succeed. God would not give us a faculty that ever leads to false belief if properly used. This leaves open the question of the requirements for right use. Clear and distinct perception that is uncorrected, untested by other faculties, must yield truth only on the assumption that its proper use does not require tests for correction. Were this assumption dispensable, Divine veracity would equally show that sense-perception, untested by other faculties, must issue in true belief! The hierarchy of the cognitive faculties is not just independent of the Divine validation of reason; it is required by it. (This explains why Descartes introduces the hierarchy prior to the proof of the truth rule and the nondeceiving God.) Descartes is thus committed to the hierarchy, whatever the merit of my case for dissimulation. A psychological doctrine grounds the priority of reason to sense-perception: reason generates irresistible beliefs, whereas sense-perception generates suppressible inclinations to believe. This asymmetry, coupled with the claim that permanence in belief is a doxastic objective, explains the normative doctrine that the proper use of sense-perception requires submitting the beliefs it generates to tests for correction by reason, but not vice versa. Absent such correction, beliefs based on the senses are at risk of impermanence because they are infirm, liable to subversion. Withstanding tests for correction strengthens inclinations to believe arising from the senses, reducing this liability. Because reason is irresistible, it need not be subjected to tests for correction in the interest of permanence in belief. (In “Circle” and “SDHP,” I take the operative psychological objective to be unshakability rather than permanence, but the argument proceeds in the same way.) The key to these results is to construe correction as fundamentally a psychological process of subverting belief. In this way, Descartes provides a nontheological foundation for the hierarchy of cognitive faculties. It is only at this stage that an appeal to Divine veracity could come into play. With the psychological account of correction and the hierarchy in place, Descartes can argue for linkages between the proper use of the cognitive faculties and securing true belief. Since God would not give us a faculty that leads to false belief if properly used, and since the proper use of reason does not require tests for subversion, whatever one clearly and distinctly perceives is true. Similarly, when the required tests by reason subvert a perceptual belief, the belief that has been corrected is false.

introduction

11

Descartes’ posture toward the truth of beliefs based on sense-perception that are not subverted by reason is a more interesting case. Perceptual beliefs are reinforced, or firmed up, insofar as they withstand tests for correction. This is as far as I take this matter in “Priority” and the papers that follow. Consider a perceptual belief that withstands all possible tests for correction. One does not clearly and distinctly perceive the perceptual belief itself, though one perhaps clearly and distinctly perceives that the belief withstands these tests. Does Divine veracity commit Descartes to the infallibility of such beliefs, to a “truth rule” for perceptual beliefs sustained by reason? Frederick Schmitt (1986) proposes an ingenious way to avoid this strong a result. On Schmitt’s account, Divine veracity requires that beliefs are true to the extent that our faculties incline us toward them. (It would be best for Descartes to restrict this alignment to beliefs arising insofar as faculties are properly used.) The irresistibility and corresponding infallibility of clear and distinct perception are a limiting case. Beliefs based on sense-perception are stronger, and hence more reliable, to the extent that they survive tests for correction. This accommodates degrees of probability, understood in frequentist terms, within Descartes’ foundationalism. Perceptual beliefs that withstand all possible tests for correction will be strong indeed, albeit falling short of irresistibility; these beliefs, though not infallible, are morally certain. These elegant results, which depend on psychological doctrines about reason, sense-perception, and their interactions, cohere beautifully with my claim that the priority of reason is at bottom psychological.

5. The Problem of the Circle (Chapters 2–4) Let us return to the full set of claims about truth (§4): the likelihood of the truth of a perceptual belief that has withstood tests for correction is proportionate to its psychological strength; a belief corrected by reason is false; and reason is infallible (the truth rule). Though securing these claims depends on the circuit through Divine veracity, leading to the traditional problem of the circle, the psychological grounding of the priority of reason suggests a solution. Descartes maintains that the conclusion of a continuous chain of clear and distinct perceptions is psychologically irresistible at the time it is reached. This extension of the claim that individual clear and distinct perceptions are irresistible facilitates a psychological response to the problem of the circle. Attending to an uninterrupted proof of the nondeceiving God leads to irresistible belief in the claims about truth. When in the grip of the Divine validation of the hierarchically ordered set of cognitive faculties, reflection on skeptical hypotheses does not shake the irresistible conviction in the claims about truth. Descartes’ contention is that any skepticism about the faculties can be rendered psychologically inert. Disabling the global skeptical hypotheses serves the interest of the permanence and unshakability of mathematical and perceptual beliefs. Beliefs in theorems based on the recollection that they have been proved are not irresistible

12

reflection and the stability of belief

and can be undermined by the deceiver hypothesis. Attending to an uninterrupted proof of a theorem results in irresistible belief at the time one completes the proof, even when one concurrently entertains the deceiver hypothesis. One cannot, however, thwart the deceiver hypothesis by simultaneously attending to continuous proofs of every previously demonstrated theorem. Descartes does not endow the enquirer with unlimited capacity for conscious processing. (It is an important question why Descartes considers some idealizations inapt.) Similarly, one cannot simultaneously rehearse the tests for correction every perceptual belief has withstood. The procedure for arriving at irresistible belief in the nondeceiving God and the truth rule systematically contains the psychologically subversive effects of skeptical hypotheses. The texts support a variety of refinements to the psychological response. It is with this explanation of Descartes’ response to the circle in hand that I find dissimulation theories much less attractive than in “Radical” and its companion 1988 paper (§3). There remains the philosophical objection that dispelling skeptical doubts by way of a question-begging argument is no better than submitting to a pill or potion that induces irresistible belief. I respond, as best I can, in “Circle.” Descartes maintains, independently of the proof of the nondeceiving God, that clear and distinct perception constitutes our best cognitive faculty because it is psychologically irresistible. Further, sense-perception leads to conflicting beliefs, and only reason can remove the incoherence. Not to engage in reason is to consign ourselves to inconsistency or to forgo senseperception and reason alike, relinquishing any hope of a comprehensive system of beliefs (cf. §3 of “Priority”). There is no pill or other belief-inducing process remotely comparable to reason, as Descartes conceives it—a faculty that often overturns perceptual beliefs, as about the shape of a stick protruding from water—in its reach and psychological strength. In assessing the interpretive force of philosophical objections to the psychological response to the circle, it is worth bearing in mind the degrees of freedom in associated accounts of the development of Descartes’ position. Perhaps, in composing the Meditations, Descartes sought from the outset to secure belief that enjoys a special psychological status. It is this version of the response that comes under maximum stress from the objections, and hence from a principle of charity. Alternatively, perhaps Descartes initially deemed truth the fundamental doxastic objective, turning in its place to unshakability, under pressure from the problem of the circle. Similarly, it is possible that any transformation in the project was not fully conscious, or that Descartes vacillated between truth and a psychological objective. These are options within the framework of the psychological response. What matters here is that interpretations that find vacillation or evolution in Descartes’ view mitigate misgivings, on grounds of charity, about attributing the psychological solution to Descartes. It is difficult to underestimate Descartes’ investment in psychological claims. The priority of reason—an indispensable component of Descartes’ epistemology, one independent of the Divine validation—is rooted in a psychological asymmetry (§4). As Schmitt shows, Descartes’ psychological doctrines permit

introduction

13

subtle discriminations in justification among perceptual beliefs (§4). The psychological construal of priority also fits nicely with interpretations that take the “necessity” of the eternal truths (§2) to consist in a compulsion to assent. If so, the irresistibility of clear and distinct perception melds Descartes’ metaphysics of necessity with his epistemology. Further, in his accounts of knowledge and its foundations, Descartes advances a cluster of interconnected psychological notions—firmness or solidity, and unshakability, among them. In “SDHP,” I argue that Descartes has his eye on a complementary, psychological conception of doubt as a state where belief is shaken. Peirce was wrong to take Descartes to task for focusing on “paper” or “verbal” doubts. Irresistibility would not preclude doubt, if doubt need not be “real and living.” Peirce overlooks passages, some in Meditations I and II, where doubt is a state that unsettles belief. On an attractive reading, Descartes anticipates the Peircean identification of doubt with an unsettled state—a conception present in Hobbes. Ironically, Descartes’ objective of unshakable belief, belief incapable of being unsettled, is closely related to Peirce’s objective of the fixation of belief, the settlement of belief in the long run. (This is not the last instance we will encounter—see §§14–15—where a figure exaggerates doctrinal differences for the sake of a contrast with a compelling philosophical moral.) The psychological response to the problem of the circle utilizes resources that are entrenched in Descartes’ thinking.

6. Truth and Stability (Chapter 4) We can locate Descartes in a tradition that assigns stability a pivotal epistemological role. This is the broader task of “SDHP.” I group Sextus and Hume, together with Descartes and Peirce, as figures for whom psychological objectives determine the epistemic status of belief. Descartes’ project of explaining scientia dictates unshakability as an objective (§1). The problem of the circle exerts pressure in the direction of construing unshakability psychologically (§5). Hume, consonant with his project of explaining everyday knowledge (§1), sets the threshold lower, at belief that is settled at the time it is held (though see §§7, 9). Assuming settled belief is tranquil, this is a special case of Sextus’ objective of ataraxia, quietude. (In “Priority,” where I began to develop the comparison between Descartes and Hume, I took both figures to focus on permanence in belief.) My overarching claim is that there are significant strands of thought in Descartes and Hume, as well as in Sextus and Peirce, that focus on stability as an objective of inquiry. It is worth noting, as Richard Popkin has shown (1960) and Edwin Curley emphasized (1978), that Pyrrhonism enjoyed a substantial revival beginning late in the sixteenth century. I attempt to defuse the objection that the objective of inquiry cannot be something other than truth. Belief, we can grant, aims at truth, in that we aim to regulate belief so that it is responsive to truth. This does not

14

reflection and the stability of belief

preclude belief’s aiming at stability, if the latter is a different and higherorder objective. Sextus represents the skeptic as initially seeking true judgments in the interest of achieving quietude. For Peirce, reflection on the method of authority—its dependence on historical accidents extraneous to the truth of the beliefs it generates—is unsettling precisely because belief aims at truth. For Descartes, stability is likewise a higher-order objective than truth. In assessing the faculties, Descartes cannot directly exploit the aim of truth, except on pain of begging the question against the skeptical doubts. (Peirce, unlike Descartes, feels free to ignore these doubts as nominal.) The higher-order objective of unshakability fills the void in the available means of assessment. In yielding to unshakability, Descartes need not relinquish the aim of truth; conclusions about the truth-conduciveness of the faculties meet the test of unshakability (§5). Only Sextus renounces the aim of truth. The Pyrrhonian sets out to achieve tranquility by way of true belief but finds himself mired in equal and opposing arguments, isostheneia. Unable to resolve the oppositions, he abandons the aim of truth in favor of suspending judgment, epoché. (Famously, the Pyrrhonian thinks suspense produces tranquility after all.) Descartes has no such need to reject a goal of truth. Descartes, Hume, and Peirce retain the aim of truth, albeit subordinated to stability. Sextus, Hume, and Peirce share an account of why stability is prized. Settled states are pleasant or satisfying; unsettled states unpleasant, aversive. Hume and Peirce apply these claims to belief and doubt. I elaborate the similarity in their views in 1995b. Hume, before Peirce, identified doubt with a wavering, unsettled state—the opposite of belief, a state that is firm or settled, because infixed. (My reading goes against the grain of Hume’s official theory that belief is a lively idea. See the next section.) Doubt is uneasy or uncomfortable, providing a motive for its own removal. Descartes can be read as identifying doubt with an unsettled state (§5), but he does not take the additional step of holding that unsettled states are uneasy. Sextus, Hume, and Peirce provide a naturalistic account of the value of stability, rooted in the affective character of unsettled states. For Hume and Peirce, this amounts to a naturalized account of epistemic obligation, as deriving from the motivational force of the felt uneasiness in unsettled doxastic conditions. Hume has criticisms of Sextus from within their shared framework. One is well-known: since some beliefs are irresistible, suspension of belief offers no general path to quietude. Another runs deeper: suspense, could it be achieved, would not constitute a pleasant condition of balance or equilibrium. Engaging equipollent, opposing arguments results in oscillation—one belief ascending and then the other, as with parties in combat or struggle, a markedly unsettled, unpleasant state. Notwithstanding these criticisms, Hume retains the Pyrrhonian objective of securing quietude and agrees that settled states are pleasant. His contention is that equipollence is unsettled and agitated; if a settled and satisfying state can by achieved, it must be within a system of belief.

introduction

15

7. Custom, Uneasiness, and Reflection (Chapters 4–5) A main line of argument for my stability interpretation of Hume focuses on passages in I.iii.9 (“Of the effects of other relations and other habits”). Hume claims that association by the relation of cause and effect, unlike association by contiguity or resemblance, produces a state that both qualifies as belief and merits epistemic approval. The claims about belief and justification, however, are run together, without Hume remarking on this fact. This unacknowledged intermingling requires explanation. Additional textual phenomena present interpretive problems exhibiting the same structure (SJHT, §§III.1, III.3). The challenge is to locate a property that is necessary for belief and also sufficient for justified belief. Stability plays this dual role, seamlessly integrating Hume’s accounts of belief and justification—hence the title of chapter 5 (“Integrating”). In the form of steadiness or fixity, stability is necessary for belief. Hume’s pronouncements about belief and states that mimic belief are consistent only if belief is a steady or infixed disposition. (Hume used forms of the verbs ‘fix’ and ‘infix’ and the adjective ‘fixt’. Had he used ‘fixity’, the noun might have appeared as an entry in the Selby-Bigge index, pointing interpreters in the direction of a dispositional account.) Hume overlays a theory of justification on this independent theory of belief. At I.iii.10 (“Of the influence of belief”), belief is nature’s provision for a steady influence on the will and hence on action. The point of epistemic distinctions is to call attention to circumstances in which the presence of conflicting beliefs undermines a belief’s influence and thereby its natural function, circumstances in which thought is susceptible to instability, even though belief has been achieved. A state might be steady, qua belief, without being stable overall, steady in its influence all things considered. The steadiness of belief, however, is sufficient for justification, other things being equal. I elaborate this and other arguments for the stability interpretation in 2004. As noted in the previous section, for Hume doubt, as the opposite of belief, is unsettled and hence uneasy. According to Hume, an uneasy disturbance results from any sudden change, as arises in the alternating manifestations of contradictory beliefs. Uneasiness also results from the discovery of a difference in view with another person (T 592–93; Essays 60–61). The direction of Hume’s thinking is to find uneasiness, an unpleasant sense of doxastic conflict, not only in inconsistency or genuine disagreement but also in any surface discrepancy or ostensible divergence among beliefs. Attributing uneasiness to a variety of disparities in doxastic states, Hume anticipates Leon Festinger’s theory of cognitive dissonance (SJHT, §I.3).9 Though chapters 5,

9. For literature on psychological discomfort in Festinger’s framework, see Olson and Stone 2005, 240–41.

16

reflection and the stability of belief

7, and 8 develop the stability interpretation without relying on this layer of Hume’s position, it is important for Hume that uneasiness bypasses reflection in motivating belief revision. Uneasiness plays a role in explaining the epistemic achievements of unreflective creatures (§1). If induction depended on argument or reflection, animals, children, and many adults could not have knowledge of the unobserved. Accordingly, “Nature must have provided some other principle [than any process of argument or reasoning], of more ready, and more general use and application” (EHU 106). Customary association by the relation of cause and effect stands in for reflection in generating belief. Likewise, nature does not leave it to reflection to discover conflicts and motivate their removal. A number of factors help explain why Hume does not call attention to this parallelism. The role of uneasiness in the economy of the mind is secondary in comparison to that of infixing. Further, even if the course of belief revision prompted by uneasiness can be explained along associationist lines, Hume has no associationist explanation of uneasiness itself. Also, since uneasiness is often removed, and stability restored, by second-order habits or general rules (§§10–11), dissonance might have seemed to Hume of little use to animals.10 Even so, identifying psychological mechanisms that contract the role of reflection is integral to Hume’s project.

8. Stability and Intense Reflection Considering the interconnections between justification, stability, and reflection at its most intense sheds light on Hume’s preferences among different formulations of a stability theory. Intense reflection uncovers a contradiction or antinomy with respect to the existence of matter, and it subverts all belief by reducing the probability of any proposition to zero (§2). In these contexts, reflection encompasses a priori reason but also resolute application of experimental reasoning. The question is what to make of these developments: do the destabilizing effects Hume delineates in I.iv.7 place in jeopardy the prospects for justification or the epistemic status of intense reflection? The best answer takes seriously my root interpretive idea that stability is the measure of justification: intense reflection, to the degree that it is destabilizing, is unjustified—so much the worse, epistemically, for intense reflection. Unrelenting reflection is one belief-forming mechanism among others—causal inference and other imaginative propensities (§§1, 10). Reflection has no antecedent privilege but must itself be subjected to the stability test for justification. This is what I ought to say and often do say.

10. For evidence of dissonance in monkeys and discussion of the possibility of mechanisms that reduce dissonance without relying on reflective capacities, see Egan, Santos, and Bloom 2007.

introduction

17

I emphasize, for example, that “reflective approval” interpretations—in the style of Annette Baier (1991) and Christine Korsgaard (1989 and 1996)—wrongly assume that, for Hume, a vantage point of keen reflection has special normative standing (SJHT, §§I.5, III.5–6, and 2004, §I.1). In earlier articles, I focus on the immediate destabilizing effects of intense reflection and infer that Hume holds that no belief is or can be justified—so much the worse for justification. This inference illicitly imports into my discussion the very prejudice—that intense reflection is epistemically privileged, and hence the perspective from which stability is rightly assessed—that Hume wants to combat. As a result, I mistakenly say or suggest—in §7 of 1990 (chapter 2), §6 of 1991, §§1–2 of 1995a (chapter 7), and §5 of 1995b—that Hume reaches the unqualifiedly destructive, negative, or pessimistic conclusion that no belief can be justified. Though I consistently portray this result as opposing a constructive intention to identify a basis for epistemic discriminations among beliefs, it is worth clarifying where I must backtrack. My 1990–95 articles implicitly attribute to Hume a demanding version of the stability theory: belief-forming mechanisms that operate in a subject are evaluated in terms of their tendency to produce states that are stable, steady in their influence, for a person who is maximally reflective. Justification requires stability under full reflection. On a less demanding version, belief-forming mechanisms that operate in a subject are evaluated in terms of their tendency to produce states that are stable, steady in their influence, for that very subject, given the degree to which he himself tends to be reflective. Justification is a matter of stability for the actual subject. Hume must favor a theory in the region of the less demanding version. As we have seen, the objective of stability motivates an unfavorable assessment of intense reflection, insofar as it is destabilizing. What is more, the stringent version of the stability theory is antithetical to Hume’s project of accounting for unreflective epistemic achievements on their own terms, not derivatively by way of their relationship to those of highly reflective subjects (§1). In “Integrating,” I distinguish the two versions and leave open the question of which we best attribute to Hume. In SJHT (§§III.5–6), I argue that the evidence favors attributing to Hume the less restrictive theory. This progression took me in the right direction. Unfortunately, the mistake of taking Hume to privilege intense reflection surfaces in another form, within the confines of this less demanding version. Consider a subject, such as Hume himself, who achieves intense reflection, destabilizing his beliefs at the time. In “Integrating” and SJHT, I take it that, in Hume’s view, no belief is justified for such highly reflective subjects (§4 of “Integrating” and SJHT, §III.6). Hume stresses that intense reflection dissipates outside the study. My inferring that for Hume no belief is justified for subjects for whom reflection, at its peak, is intense and destabilizing recapitulates my error in the early papers; a stability theory ought not presuppose any privilege either for the intense reflection of a fully reflective subject or for the peak reflectiveness of actual subjects. The latter presupposition enforces the result that the less

18

reflection and the stability of belief

demanding theory, applied to highly reflective subjects, yields the same negative assessment of justification as the demanding theory. There are alternatives—brought to my attention by Schmitt (2004, §3)— to tethering justification to peak reflection. A temporal stability theory takes its cue from the fact that destabilizing peaks of intense reflection are not sustained; it assesses stability within a temporal segment of a subject’s beliefs. Apart from periods of intense reflection, justification can remain fully intact. An average stability theory takes its cue from the fact that periods of intense reflection are brief; it assesses stability in terms of the (weighted) average stability of beliefs over time. A subject’s faculties can tend to a high degree of average stability, yielding justification even within periods of intense reflection. Either alternative suffices to explain how Hume makes sense of carrying on with the science of man in Books II and III, in the aftermath of the destabilization in I.iv.7. The temporal and average stability theories preserve justification by giving epistemic credit to stability when reflection is not intense (my 2004, §§II.3 and IV.1). On either option, destabilizing intense reflection reduces justification: it either destroys justification during intense reflection, as on the temporal stability theory, or it lowers the average stability, and hence the level of justification, for a subject’s total doxastic output over time. This result carries Hume’s broadside against reflection to another front. Reflection is not capable of providing Cartesian foundations for knowledge. Besides, many creatures possess knowledge and improve their epistemic position but do not much reflect. In addition, overly assiduous reflection spells trouble, generating destabilizing contradictions and subverting belief. Michael Williams has a lovely development of John Passmore’s idea that the destabilizing effects are short-term; in the longer run, they are chastening, eventuating in tempered reliance on intense reflection and on abstract and speculative theory. (See the final sections of chapters 11 and 12 and my 2004, §IV.3.) I believe this is most true to the Enquiry, though there are hints in this direction in the Treatise (Williams 2004, §6, and Schmitt 2004, 307). Importantly, any compensating or offsetting effects of intense reflection do not take the form of refining its results—for example, by locating a distinction that dissolves the contradiction about the existence of matter. Stability is enhanced only by reining in reflection. Hume’s indictment of reason and reflection, as conceived in the Cartesian tradition, depends on arguments in I.iv.1 and I.iv.4, as exploited in I.iv.7, that spin out of control (§2). Hume’s failure to reevaluate them is evidence of his youthful glee in their results (cf. SJHT, passim, and 2004, §§IV.1–3). Though Hume is overzealous, he wants to call into question the blithe Cartesian assumption (§2 of chapter 2) that reason is consistent and transparent. From this perspective, he could have welcomed non-Euclidean geometries, intuitionist logics, and alternative axiomatizations of set theory, not to mention the set-theoretic paradoxes. In Treatise I.ii, reason is saddled with paradoxes relating to infinite divisibility and geometry removed from its domain (cf. Fogelin 1985, ch. 2). Reflection itself suffers indignities,

introduction

19

uncovered in its own exercise or through empirical investigation of its limitations. This is the significance of the first (and least problematic) step of the argument in I.iv.1, where Hume seeks to bring demonstration under empirical control in light of its fallibility in dealing with intricate questions. I return to these themes in the final section.

9. The Less Demanding Stability Theory and Easy Justification Does the view that justification depends on the tendency of mechanisms to produce stable belief for the actual subject make justification too easy? This objection is more worrisome than that directed against my interpretation of Descartes, that epistemic authority cannot derive from the irresistibility of clear and distinct perception (§5). Whereas a pill producing irresistible belief is a philosopher’s fancy, there do exist processes (such as indoctrination) that infix belief and do not earn our epistemic esteem. That settled belief, unlike Cartesian unshakability, is well within reach exacerbates the problem. The difficulty is inherent in the stability tradition. Hume would not want to fall back on Peirce’s response, settlement of belief in the long run (my 2004, §IV.4); however, he has available a more powerful version of Peirce’s argument against the method of authority (§6). For Peirce, unsettling that method requires reflection on the character of the contingencies that give rise to the beliefs it generates. Hume’s psychological doctrines contemplate a prereflective surrogate for the role of reflection. Merely encountering apparent disagreement with another person shocks or rattles our beliefs, causing disturbance (§7). (In the stability framework, this observation bears on the epistemology of disagreement, with epistemic peers and otherwise.) Since one is likely to encounter differing judgments, there is a tendency for beliefs based on authority to become unsettled. Similarly, beliefs based on authority and dogmatism will tend to conflict with causal inference. These are instances of a general strategy (SJHT, §III.6, and my 2004, §IV.4). The less demanding theory requires of belief-forming mechanisms a tendency to produce stable belief. Similarly, natural virtue is assessed—think “virtue in rags” (T 584)—in terms of a character trait’s tendency to benefit society. Even in the moral theory, where the notion of a tendency is frequently invoked in III.iii.1 and III.iii.3, it is a placeholder for a more determinate theory. The normative implications of Hume’s epistemology, once “tendencies” are introduced, are no less fuzzy; he need not be saddled with any simple inference from the fact of settled belief to a conclusion about justification. Justification can also seem too easy where subjects ignore relevant information. Here, too, Hume might appeal to tendencies. Alternatively, he could add a proviso within the less demanding theory (a comparative stability account): there is no alternative belief-forming mechanism that is available to the subject, has at least as great a tendency to produce stable belief, and would lead the subject not to hold the belief. “Availability” would be characterized

20

reflection and the stability of belief

narrowly, in keeping with the spirit of the requirement of stability for the actual subject, however reflective (SJHT, §III.6). The less stringent stability theory can thus condemn ignoring readily accessible information, including that others disagree. In sum, a variety of refinements and epicycles—albeit not ones Hume bothered to pursue—can blunt the force of the objection from easy justification.

10. Case Studies: Semantically Defective Beliefs, Accidental Conjunctions, Variation in Confidence (Chapters 6–7) Chapters 6–8 illustrate the capacity of the stability interpretation to illuminate some main interpretive issues. (I defer discussion of chapter 8 to §11.) In turning his attention to nondemonstrative psychological transitions that result in belief (§1), Hume, an acute psychological observer, introduces several important propensities other than customary inference about unobserved causes or effects. These include mistaking merely resembling for strictly identical objects and adding a new relation to related objects. (Humean “projection”—the mind’s spreading itself on external objects, transferring an internal feeling to its external cause—is a special case of this propensity to complete a union.) The two propensities are essential to Hume’s explanation of metaphysical beliefs—in material substance, souls, body, and robust necessity (not reducible to constant conjunction). Hume needs grounds for allocating the nondemonstrative transitions not founded on causal inference either to the understanding or to the narrow imagination (§1). In “Hume’s Explanations of Meaningless Beliefs,” I argue that the propensities to ascribe identity and to complete a union contribute to inclinations to hold inconsistent beliefs, giving rise to attendant uneasiness (§7). Metaphysical beliefs constitute attempts to relieve this discomfort by obscuring the conflicts. Though more needs to be said about the epistemology of this process (cf. SJHT, §V.4), the propensities give rise to an unstable outcome, conflicted attempts to relieve uneasiness. I also tackle a prior question. Hume often declares ‘substance’ (or ‘substratum’), ‘external object’, and (robust) ‘necessity’ meaningless, suggesting that one cannot possess the beliefs Hume seeks to explain.11 I contend that in I.iii.14, I.iv.2–3, and I.iv.5–6, Hume implicitly relaxes the Lockean strictures on meaning advanced in I.i. A kind of content, implicated in the attempt to resolve the imaginative conflicts, accrues to the suspect metaphysical concepts. The most promising hypothesis is that Hume began the Treatise as a Lockean about meaning but worked his way beyond the confines of that position. Further, Hume’s invoking psychological disquiet as he homed in on explanations of metaphysical beliefs was a prominent departure from 11. Of these cases, the discussion of necessary connection is not incorporated into SJHT.

introduction

21

associationism (§7). Hume would have better signposted his views by revising I.i, both to explain how there can be beliefs that lack Lockean content and to preview the range of psychological resources admissible in the science of human nature. Even Hume’s favored nondemonstrative transitions, those underpinning memory and causal inference, give rise to beliefs that are “unphilosophical.” I address these beliefs in “Hume on Stability, Justification, and Unphilosophical Probability.” In I.iii.13, Hume turns his attention to two main classes of phenomena: variation in degree of confidence as memory fades and like cases (Hume’s first three kinds of unphilosophical probability) and beliefs due to observed conjunctions that are accidental (the fourth kind). “Of unphilosophical probability” provides evidence of Hume’s sensitivity to the epistemology of accidental uniformities. Granted, Hume focuses on sets of conditions that contain superfluous or accidental members (A) as well as genuine causes (C). We can suppose the set C alone constitutes a minimal sufficient condition for a later event or state of affairs E; the conjunction between A-and-C and E piggybacks on an underlying, lawful regularity between C and E. Early in the Treatise, Hume contemplates a chance regularity that does not arise by piggybacking (T 4). How could Hume attend—in I.iii.8 (104–5) and I.iii.15 (175), as well as I.iii.13—to weeding out beliefs based on accidental conjunctions, while failing to address such conjunctions as counterexamples to his regularity analysis of causation? This is an open question in Hume scholarship. One of those little-read sections, I.iii.13, is also crucial to an appreciation of Hume’s epistemology and contains considerable evidence for a stability interpretation. Beliefs based on observing accidental conjunctions give rise to a special case of the problem of “inductive inconsistencies.” These are not formal contradictions but rather different probability estimates relative to different bodies of evidence—in I.iii.13, relative to accidental and nonaccidental generalizations. Variation in degree of confidence is an instance of the psychological biases familiar from the work of Kahneman, Slovic, and Tversky (1982). (Perhaps Hume’s associationist embellishment has discouraged reader interest.) Why regard the variation as reflecting “bias,” “distortion”? Hume’s answer is that memory and causal inference win approval overall, on grounds of stability, but their operation results in more localized instabilities. I suggest that reflection on the second-order belief that confidence in memory declines over time is destabilizing. I replicate this account in SJHT, §IV.3, but also improve on it in §IV.5. There is no contradiction in declining confidence in memory over time. Though the changes in the degree of confidence are gradual, they accumulate to large differences. If one so much as notices this variance, the general principle that apparent disparities are unsettling and give uneasiness (§7) comes into play in application to doxastic states over time.

22

reflection and the stability of belief

The instabilities reviewed in I.iii.13 are eliminable in light of the availability of general rules that function as stabilizing devices.12 General rules are second-order beliefs, habits that are themselves rooted in perception, memory, and custom. For Hume, as Williams emphasizes (his 2004, §6), these faculties are self-correcting, when more expansively employed to examine classes of beliefs. (By contrast, intense reflection in the Enquiry acts to contain its own use. See §8 here.) This is another narrowing of reason’s portfolio; for Descartes, the task of correction, even correcting sensory illusions and perceptual relativity (§3 of chapter 2), falls to reason.

11. Moral Judgment (Chapter 8) Hume’s theory of moral judgment is propitious territory for the stability interpretation: “In order . . . to prevent those continual contradictions, . . . we fix on some steady and general points of view” (T 581–82, Hume’s emphasis). The “contradictions” are the intrapersonal and interpersonal differences in judgments of virtue and vice due to variability in the intensity (and maybe the valence) of sentiments produced by sympathy. Much as confidence in memory degrades with distance in time, the intensity of feelings arising from sympathy with pleasures and pains of others increases with psychological closeness—temporal and spatial and other special relations (blood and family, friendship, acquaintance, nationality, language, profession). Why regard the variation in sentiment as objectionable, rather than just a reflection of different points of view? The quotation provides the key. The variation in judgments of virtue, much as that in confidence in memory, calls for a steadying device. Judgments corrected in accordance with general rules are normative because they result from stabilizing mechanisms. Hume can appeal to the uneasiness in unsettled states to explain how fluctuation in judgment motivates adoption of a stable point of view. While there is no contradiction in judgments of virtue from different psychological perspectives, the interpersonal variation can be pronounced—the agent judged by friend and foe (T 472)—and the intrapersonal variation sudden, when a stranger becomes an acquaintance (581). Such outward disparities are uncomfortable, disconcerting (§7). Pressure toward a stabilizing point of view does not depend on the presence of unvarnished contradictions or even on reflection on the variation in judgment. Hume is undertaking to explain psychological pressures operating on the common person, the vulgar. Reflection can enhance or detract from the stability of moral judgment, but it is not required either to motivate efforts to adopt, or successfully to secure, the privileged point of view. I discuss this much,

12. In the case of the first three kinds of unphilosophical probability, I suggest a different version of this solution in chapter 8, note 35 and SJHT, §IV.6, than at the close of §4 of chapter 7.

introduction

23

including the analogy to unphilosophical probability, in SJHT, §§IV.5, and 2004, §§I.1 and I.4.13 In “Hume’s Agent-centered Sentimentalism” (“Sentimentalism”), I turn to a substantive characterization of the steady point of view. Moral philosophers often read Hume as a utilitarian; the moral judge sympathizes equally with the pleasures and pains of everyone. If the judicious spectator thus abrogated sympathy’s intensification of special relationships, moral judgment would hardly be grounded in sentiment. In spontaneous, uncorrected evaluations, one adopts one’s own point of view (as friend or enemy of the agent, as stranger, as near or remote). The resulting differences in interpersonal evaluations (and in intrapersonal evaluations over time) shake us from this perspective. On which point of view does one settle down? In the special case of evaluating one’s own character, sympathy spontaneously gives the greatest weight to oneself and one’s family and friends. Were others evaluated by that standard, centered on oneself, one would encounter unsettling variation in judgment. The remedy is not to nullify the psychology of sympathy, but to transfer the spontaneous perspective in self-evaluation to the agent, indexing sympathy’s operation to the circle of the agent under evaluation. Because one sympathizes with everyone, this outcome is universalist. Because one sympathizes most intensely with those closest in psychological space, adopting the sympathetic perspective of the agent’s circle weights most heavily the pleasures and pains of the agent and his family and friends. Hume’s moral theory requires an agent’s partiality to his circle. It also generates principles along the lines of agent-centered restrictions in utilitarian theories. Hume reaches results similar in content to a moral theory based on differential loyalties—to family or tribe, country or nation—advanced by Andrew Oldenquist (1982). Much as Hume’s construction of the steady and general point of view does not require sympathizing equally with everyone, there will be no requirement—more suitable for a Cartesian moral inquirer—of full information. I argue in 2004 (§I.3) that a requirement of incorporating readily available information (cf. §9 here) is at once salient and suitably stabilizing. Hume’s steady and general point of view is standardized to achieve stability but is far from idealized in the manner of “ideal observer” theories. These implications mark a significant contrast with Hume’s theory of induction. Neither basic inductive inference nor moral judgment results from reflection, much less from demonstrative reason. Even so, first- and higher-order habits stabilize inductive beliefs whose content accords well with those a rationalistic account of induction would seek to authorize. The effect of custom, in many cases, replicates an inductive straight rule. In ethics, the demand for stability, coupled with Hume’s assumptions about 13. SJHT, however, might be viewed as advancing two different accounts, analogous to the two accounts of instabilities in the context of unphilosophical probability (§10). In §IV.4, reflection on second-order beliefs about the variation is destabilizing; in §IV.5, uneasiness, a direct product of the variation, motivates change in judgment.

24

reflection and the stability of belief

psychological salience, leads to more contentious implications. “Sentimentalism” pushes this divergence further: in a thoroughgoing development of Hume’s position, the judicious spectator adopts the point of view of the agent, rather than that of the agent’s circle. This result delivers the prescriptions of “self-referential altruism” (altruism weighted toward oneself and those with whom one has special relationships), a theory outlined by C. D. Broad (1942, 43–57, and 1953, 277–82). For Hume, any special resemblance intensifies sympathy. Why does he give prominence to sympathizing with family, friends, acquaintances, and countrymen, not coreligionists or members of the same trade? (I address this question in 2004, §I.3.) Oldenquist, though not engaged in Hume interpretation, observes that some groups to whom one bears special relationships are nested (1982, 178, 180–81). The groups family, friends, acquaintances, countrymen, and humans are (or, in Hume’s time, were) for the most part ordered by set-inclusion. In Hume’s psychology, these groups also generate progressively less sympathetic intensity. This psychological ordering coincides with the nesting by set-inclusion, rendering the enumerated special relationships especially salient. For the purposes of the most basic judgments about natural virtue, it is the relationships of family, friends, and so forth, rather than sharing a surname, profession, or religion, that operate as sympathetic templates or screens within the steady and general point of view.

12. Kemp Smith on Books I and III of the Treatise (Chapter 9) According to Norman Kemp Smith, Hume was primarily interested in ethics, working out, and composing, his moral theory prior to his theory of the understanding. (Kemp Smith’s 1918 “patchwork” interpretation of Kant’s first Critique likewise offered a hypothesis about manner of composition.) Further, the moral theory determines a parallel theory of judgment. Since morality is solely a matter of sentiment, factual judgment, too, is thoroughly subordinated to feeling. Oddly, as I discuss in “What Is Worth Preserving in the Kemp Smith Interpretation of Hume?” there is little or no hint in Kemp Smith of the role of corrections to sentiment by general rules in constituting the standard of merit or virtue. Perhaps because his principal philosophical interest was the understanding, not morals, Kemp Smith’s tendency is to reduce Hume’s moral theory to a slogan, that morality is a matter of feeling. The version of Kemp Smith’s interpretation of Hume on the understanding that circulates in the literature is similarly reductive, emphasizing the thorough subordination of reason to feeling and the seeming implication that factual belief is arational or nonrational. James Ward Smith, reading Kemp Smith’s 1941 book in this way in a 1960 article in the Philosophical Review, built a muscular case that the book’s interpretation, so construed, is mistaken. It is dismaying that Smith’s article received scanty attention; there

introduction

25

is no trace of it in the literature after 1971, even in books where the reductive version of Kemp Smith’s view is front and center. Kemp Smith’s book contains another strand, apart from the one Smith refuted. Many commentators rely on Kemp Smith’s manifesto in Mind, ignoring the distinctive contributions of his enormous volume published thirty-six years later. In the book, there is the recognition that Hume wants to distinguish, within “experience,” among good and bad customs. Not all beliefs based on memory and causal inference are equally legitimate. It is this insight—not the thesis that belief is subordinate to sentiment—that genuinely illuminates Hume’s project in Book I. Kemp Smith’s commitment to idealism has obscured the best elements in his interpretation. On the one hand, he saw that Hume offers a “naturalistic” account not just of belief, but of epistemic distinctions among beliefs within the faculty of the understanding. On the other hand, Hume’s resting his account on instinctive propensities was anathema to Kemp Smith. His idealism took intellectual and moral values to be present from the beginning of the universe, determining its structure. For Kemp Smith, Hume’s rooting values in evolved human nature was ill-conceived, so much so that he distanced himself from his best interpretive insight. (This contamination of his interpretive work would have appalled Kemp Smith.) In taking Hume to discriminate among customs, Kemp Smith has his eye, in part, on phenomena treated in I.iii.13. Nonetheless, much as he ignores the stabilizing role of general rules in Book III, Kemp Smith pays no serious attention to their role in smoothing out variation in judgment in I.iii.13 (§10). It is easy to lose sight of the fact that it took Hearn’s articles in the 1970s and Brand’s 1992 monograph to bring an analogy between general rules in Books I and III into full view. In any case, Kemp Smith does not consider the analogy’s bearing on his central interpretive claims. For example, in Book III Hume likens the use of general rules to correcting perceptual judgments of size for variation in spatial distance (T 582, 603). This comparison does not appear in Book I, save in the Appendix (632/1.3.10.12)—published with Books II and III, after Book I. Had Book III been composed first, we would expect to find the size example incorporated into Book I proper, not added as an afterthought. Putting order of composition to the side, suppose Kemp Smith had paid heed to corrections and general rules in Book III, to Hume’s seeking to discriminate among judgments of value. This might have led to a better appreciation of the analogous developments in Book I, in turn helping Kemp Smith to overcome his idealist misgivings about attributing to Hume a naturalistic account of epistemic distinctions.

13. Internalist and Externalist Versions of Hume’s Epistemology (Chapters 5 and 10) My arguments in favor of a stability interpretation are heavily bottom-up, extracting from textual detail. Proceeding top-down, from my large-scale

26

reflection and the stability of belief

thesis about Hume’s project, we can abstract to a broader class of epistemologies that might plausibly be attributed to him. Accounting directly for the epistemic success of ordinary creatures, however reflective, constrains the epistemology Hume might fashion (§1). In my favored version of the stability theory (§§8–9), justification depends on the tendency of beliefforming mechanisms to produce stable beliefs for the actual subject. Epistemic assessments of the beliefs of natural philosophers, the vulgar, children, and animals fall out directly as special cases. The theory shares this power or generality with reliabilist and proper function theories. We need only suppose operative mechanisms that are reliable, or properly functioning, and that do not require reflection. All these theories are externalist, taking justification to depend, at least in part, on features of the mechanisms that produce beliefs—facts “external” to a subject’s belief system. Hume’s constraint is also compatible with a specialized internalist theory. To isolate this option, it helps to return to stability’s dual role in Hume’s theories of belief and justification: a belief is justified, other things being equal, provided it results from a mechanism that tends to produce states that are settled, qua belief (§7). “Other things being equal” is included because the justification is prima facie, that is, defeasible; the operation of other mechanisms could tend to produce conflicts that render beliefs unsteady in their influence, undermining justification overall. The products of belief-forming mechanisms are defeasibly justified, simply in virtue of resulting from a mechanism that tends to produce belief. We can transmute this theory into an internalist variant that appraises beliefs independently of mechanisms: a belief is defeasibly justified, simply in virtue of being settled and infixed, simply in virtue of being a belief. Belief has defeasible justification intrinsically.14 In a positive coherence theory, some nontrivial, positive degree of coherence is necessary for justification. Though Hume had Cartesian or foundationalist theories in his sights, his constraint excludes positive coherence theories as well. The knowledge of animals, children, and unreflective adults should no more depend on the ideal of a maximally coherent belief system than on that of a fully articulated foundationalist structure. A coherence theory can lower its threshold for justification, but if the requirement is a degree of explanatory or probabilistic coherence that only highly reflective cognizers can achieve, others will again be consigned to a derivative or second-rate status. A reflective approval theory (§8) encounters the same problem. In the internalist stability theory, incoherence functions negatively, as a defeater that undermines or overrides the defeasible justification that automatically accrues to every belief.15 This negative coherence theory—just as 14. The two theories agree in their assessments of defeasible justification, except in two cases: uncaused beliefs, not resulting from any mechanism, and beliefs that result from mechanisms that do not regularly produce belief. 15. Both negative and positive coherence theories deny relations of epistemic priority (in a sense that generalizes the notion at §7 of chapter 11) among classes of beliefs, but the negative theory adds that every belief possesses (prima facie) justification. Harman’s “general foundations”

introduction

27

well as the externalist theories—directly explains the justification of unreflective creatures. Hume’s doctrine that a feeling of uneasiness often does duty for reflection (§§7, 10, 11) can add a helpful twist: felt conflicts qualify as “incoherence.” (In developing the interpretation, “Integrating” does not rely on this Humean resource.) To the extent that explanatory gaps, probabilistic inconsistencies, and contradictions manifest themselves in discomfort that motivates its own removal, justification can be defeated without reflection on the relations among beliefs that give rise to these forms of incoherence. Hume’s constraint on an epistemological theory leads to either an externalist or a negative coherence theory. Both countenance justification even absent the argument or reflection required to produce a sophisticated structure (whether vertical or otherwise) of interlocking beliefs. It is because they share this feature that the epistemologies in view are vulnerable to a common objection, that they make it too easy to possess justification or knowledge (§9).16 This complaint, often leveled against externalism, inevitably awaits epistemologies that allow unreflective knowledge that is not deemed second-class.17 I do favor the externalist stability reading; the evidence, as at the first paragraph of I.iv.4 (§1), is that Hume prefers to evaluate beliefs in terms of “principles” or belief-forming mechanisms, in terms of underlying intellectual character.18 This and other externalist interpretations have resources for disarming passages that invite an internalist reading: “A wise man . . . proportions his belief to the evidence” (EHU 110); “philosophical decisions are nothing but the reflections of common life, methodized and corrected” (162). Such passages contribute to the case that Hume is no inductive skeptic (§1) but do not show that he is an internalist. Externalist epistemologies do not deny that facts internal to one’s belief system are relevant to justification (cf. my 2008, 115–17). Rather than rejecting internalist intuitions and precepts, externalism seeks to explain them on its own principles.19 Plausibly, codification or systematization of belief is adaptive, reliable, and (as Hume asserts at DNR I, 134) conducive to stability.20 (Reflection that is too intense may view is a recent example (2001, 2003, 2010). Negative coherence theories can arise in retreat from pressures on positive coherence theories (cf. Firth 1964, §4) but also resonate with intuitions motivating Neurath’s ship metaphor (Neurath 1932 and Quine 1969, §1) and epistemological conservatism. For a useful (critical) overview of conservatism, see Christensen 1994. 16. I elide the question of how to factor out justification from knowledge, apart from requiring truth only for knowledge. I doubt that Hume has views on this matter. 17. For the problem in up-to-date guise, see the “easy knowledge” literature descending from Cohen 2002. 18. SJHT and my articles, with the exception of “Integrating,” advance an externalist stability interpretation. 19. Externalist criticism of Sosa’s distinction (introduced in 1985 and developed in much subsequent work) between animal and reflective knowledge is relevant here. See Goldman 2004, 86–90, and Kornblith 2004. 20. In the Treatise, however, the point of the “very dangerous dilemma” (T 267), an outgrowth of the reduction of all probability to zero (§2), is that epistemic principles themselves resist systematization. Cf. my 2004, 386–88.

28

reflection and the stability of belief

be an exception.) Externalists attempt to explain the importance of weighing evidence, insofar as such reflection is well within a subject’s reach, along similar lines (§9). Externalism generates no analogous “internalist” requirements for young children or animals. There is an obstacle from another quarter specifically to reliabilist interpretations. A reliable belief-forming mechanism tends to produce true belief. It is doubtful that Hume has a substantive account of the truth of moral judgment. Identifying true moral judgments with those adopted in the steady and general point of view (§11) would trivialize the reliability of sympathy as corrected by general rules. For a unified externalist interpretation of Hume on factual and moral judgment, we must turn to stability or proper function theories. The latter is an option, if formulated at a level of generality that does not require truth-conduciveness. It is consistent with this that proper functioning requires reliability (when mechanisms function nondefectively, in an appropriate environment) in the case of the understanding. Because extant proper function readings of Hume on factual judgment are quite rudimentary, adjudicating between stability and proper function interpretations lies in the future. I doubt that there will prove to be a case for a confident judgment.

14. Perceptual Knowledge in British Empiricism, the First Enquiry, and Reid (Chapters 11–12) Hume and Reid alike have been interpreted in terms of negative coherence, reliabilist, and proper function theories. This is a symptom of their common outlook: mundane knowledge cannot rely essentially on argument or reflection; in their stead, it must be attributed to the human and animal constitution, to instinct. Reid was far from interpreting Hume in this way. An unsympathetic critic, focused on areas of disagreement (see the next section), he was capable of misrepresenting predecessors.21 In the present instance, Reid uses Hume much as Peirce uses Descartes (§5), as a foil or counterpoint. Reid is extravagant in his inventory of fundamental instinctive faculties; he includes introspection, perception of the material world, memory, and causal inference, as well as faculties that underwrite belief in testimony, other minds, and God. I consider Hume on this topic in “Locke and British Empiricism” (“Empiricism”) and “Hume and Reid.” The first paper surveys the course of epistemology from Locke, to Hume’s Treatise, then to his Enquiry, 21. Reid overlooks even basic Cartesian doctrine about corrections to sense-perception (§4). He cites “the veracity of our external senses” as a point “wherein I cannot reconcile him to himself.” Descartes “seems to waver”: “Sometimes, from the perfection of the Deity, . . . he infers, that our senses . . . cannot be fallacious”; “At other times, . . . Descartes . . . mak[es] frequent complaints . . . of the fallacies of sense” (EIP 122–24). The complaints are about the senses when uncorrected.

introduction

29

and on to Reid, with some attention to Mill and later Anglo-American philosophy. The piece revisits issues broached in From Descartes to Hume about the status of British Empiricism as a school or movement. “Empiricism” attempts to fashion a narrative exhibiting an interesting structure, without overly falling in with simplified history. I still take the view that Berkeley’s system is an offshoot of Descartes and Malebranche, though Berkeley relies on Lockean empiricism about meaning when it suits his metaphysical purposes. Hume is less a development of Locke than a serendipitous reaction to discrete stretches of the Essay (notably II.xxi.29, II.xxxiii, IV.iii.6, and IV.xi.9–11) that stimulated his thinking in particular directions—against the background of the rationalist currents that Hume rejects. His use of meaning empiricism works at cross-purposes to his ambition to explain belief (§10). In this latter enterprise, the Hume of the Enquiry moves ever closer to Reid. The structure I locate revolves around knowledge of the external world within a foundationalist or quasi-foundationalist framework. There is room to question whether Locke subscribed to a foundationalist picture of perceptual knowledge. Some contend he was either a semantic externalist, holding that the existence of the external world is a necessary condition for the meaningfulness of perceptual beliefs, or a reliabilist. In either case, perceptual knowledge is not derived from a classical foundation of incorrigible, introspective beliefs about how things appear. Even if Locke did anticipate reliabilism, the Scottish figures arrived there by a different route. Locke was concerned to deny brutes abstract, general ideas (Essay II.xi.10–11), not to emphasize their cognitive similarity to humans. And unlike Hume and Reid, Locke had some sympathy for invoking inference to the best explanation to account for knowledge of the external world. Hume’s account of perceptual knowledge in the Treatise has its foundationalist trappings. Introspective beliefs are held to be certain, incorrigible— though even Reid, no friend of Cartesian foundationalism, held introspection, beliefs of “consciousness,” to be certain (EIP 66, 470, 519). There is the commitment to the “way of ideas,” the reification of perceptions as the immediate or direct objects of experience—a doctrine often conjoined with foundationalism. And in I.iv.2, Hume is committed to deriving knowledge of the external world from introspective beliefs about impressions. For all that, the derivation (putting aside the role of “constancy”) is rooted in custom or habit enhanced by a mental “galley.” No necessary truths license the principles of inference and hence the inductive transitions (§1); the full classical foundationalist program has already been abandoned. Section I.iv.2 can look like a halfway house between foundationalist and nonfoundationalist positions. Perhaps Hume remains in the grip of a foundationalist picture of perceptual knowledge, while rejecting a classical foundationalist account of inductive (and memory) knowledge. After all, he devotes nearly the whole of Part iii of Book I to the mechanics of causal inference, his dominant theoretical interest. Alternatively, we can think of Hume as liberated from

30

reflection and the stability of belief

foundationalism, but seeking to rely on a minimum of basic or underived empirical faculties—introspection, memory, and causal inference. One need not be a foundationalist to place a premium on simplicity. I conjecture that we best picture Hume, in this youthful work, as adopting (with varying degrees of self-consciousness) a number of starting points—the Lockean theory of meaningfulness (§10), associationism and the occurrentist account of belief (§§6–7, 10), and even aspects of a foundationalist picture. He subsequently frees himself of these straitjackets, as he follows unforeseen (or not fully foreseen) developments and turns in his argument. It is possible Hume did not much appreciate the distance between his initial commitments and his later results. At any rate, it was not his manner to revise early statements of his position (§2). In Section XII of the Enquiry, Hume conspicuously alters his stance on perceptual knowledge, allowing that belief in the external world is itself instinctive and not derived from other faculties. This is not Reid—not his host of primitive instincts, not even his instinctive propensity to believe in the extended world. Still, in the Enquiry, we have a Reidian expansion of the basic instincts beyond introspection, memory, and causal inference to include perception. Hume’s acceptance of an instinctive belief in external objects also seems to commit him to an innate idea of externality—not, as in a position Locke attacked, to explain the possibility of a priori knowledge of necessary truths, but as a condition for instinctive knowledge of empirical propositions. What are we to make of Hume’s movement in Reid’s direction? We might interpret the Enquiry, as its detractors suggest, as a simplified version of the Treatise, favoring literary merit over philosophical and psychological complexity. Or we might interpret it, as Hume instructed in the advertisement to the 1776 edition, “as containing his philosophical sentiments and principles.” Of the many differences between the Treatise and the Enquiry that might be brought to bear on this issue (cf. my 2004, §IV.3), two are of special relevance. First, while there is clear evidence in the Treatise that Hume thinks argument and reflection cannot be essential to knowledge (§6 of “Psychology”), in the Enquiry—the final paragraph of Section IV, the final two paragraphs of V, and IX in its entirety—Hume more persistently and emphatically emphasizes the acquisition of knowledge in children and animals. Whatever the admixture of foundationalist and externalist tendencies in the Treatise, in the Enquiry Hume much more squarely confronts this chief source of pressure toward externalism. It is no coincidence that the text that best supports a reliabilist interpretation (§13)—Hume’s assertion of a “preestablished harmony between the course of nature and the succession of our ideas”—appears in the Enquiry (EHU 54). Second, any derivation of belief in the external world from introspective beliefs must rely on some broadly inductive or probable argument. In the Treatise I.iv.2 discussion of the “coherence” of impressions, Hume uncovers profound problems for relying on enumerative induction (see my 2008,

introduction

31

121–22, as well as “Empiricism”) to support inference to unobserved perceptions—the “bodies” of neutral monism. Enumerativism, his favored inductive program, buckled at his own pen. Hume’s response within I.iv.2 is controversial (cf. SJHT, §VI.3), but it is natural to suppose that the section’s tortuous maneuvers came to weigh on him more and more. The alternative inductive strategy for grounding belief in the external world, theoretical induction or inference to the best explanation, comes under fire in Enquiry XII (and in passing at Treatise 212 and 216). Hume, unlike Locke, puzzled his way through the problem of deriving the belief in body inductively from introspective beliefs. He probably became increasingly convinced that belief in the external world does not admit of inductive grounding, and genuinely shifted ground in favor of a primitive instinct. The difficulties Hume found in enumerativist strategies for securing perceptual knowledge were lost on Reid, and even Mill, and not recovered until the twentieth century. Reid does play out and augment Hume’s case against theoretical inference (EIP 47–52, 76–87). Reid appreciates the appeal of inference to the best explanation as a strategy for justifying belief, whether in the external world, other minds, or the existence of God. He took his assault on theoretical inference to leave belief in these and other realms under the control of disparate, primitive instincts. The Cartesian inventory of cognitive faculties (§1), expanded in the Treatise to include causal inference, shatters into pieces.

15. Reflection and Instinct in Hume and Reid (Chapters 11–12) I intend chapter 12 as an antidote to Reid’s preferred picture of Hume as a skeptic. There were many differences to absorb Reid’s attention. He is much more hostile than Hume to inference to the best explanation in natural philosophy, has a special distaste for Hume’s unifying associationist hypotheses, and proliferates basic instincts. Whereas Reid welcomes innatism, Hume must have arrived grudgingly at the doorstep of an innate concept of externality (§14). In Reid, a Divine providentialism operates in the background. And Hume is committed to reification (§14), Reid’s master villain (though this difference is something of a red herring). Only one factor contributing to Reid’s misunderstanding of Hume will occupy us here. Hume’s goal, pursued much too enthusiastically, was to bring reflection down a good peg or two (§8). Reid agreed that reflection cannot supply a foundation for knowledge and that paradigmatic cases of knowledge are unreflective. Hume also claimed that reason is a party to antinomies and is self-subverting (§§2, 8). These pronouncements agitated Reid in the extreme. He took his providentialism to imply that the cognitive faculties, when properly used, do not significantly conflict, individually or collectively. Reid did not pause to question Hume’s seriousness in his most radical claims

32

reflection and the stability of belief

about intense reflection, much less whether any fundamental Humean doctrines commit him to the subversion of reason. There is some irony here. Reid maintains that on occasions where reason does conflict with another faculty, “a man of common sense may fairly reject the conclusion, without being able to shew the error of the reasoning that led to it” (EIP 434). All cognitive faculties are subject to error—susceptible to damage, malfunction, or application beyond their proper domain (cf. 241–52). Additional factors— preference for familiar analogies, love of simplicity, deference to authority, and passion, among others (cf. 527–41)—can pervert reason. Instinctive mechanisms, operating independently of the “weighing of arguments” (452), independently of true or “false reasoning” (467), are insulated from reason’s distinctive shortcomings. In cases of conflict, Reid concludes, there is a presumption in favor of instinct. In the Treatise, sense-perception and induction jointly correct themselves, without the aid of reason (§§10, 12). In the Enquiry, Hume goes further, anticipating Reid’s position about conflicts with reason. Instinct could not explain unreflective knowledge, if its operation were dependent on reason or argument. “[R]easoning and argumentation” is an “uncertain process” (EHU 106), “extremely liable to error and mistake” (55). Instinct is a “mechanical” (55, 108) principle and, to that extent, immune to the special infirmities that infect reason. These are the building blocks of Reid’s position, that instinctive belief trumps reason in cases of conflict. This result stands on its head the Cartesian hierarchy of faculties. Reason now plays a subordinate role in an ensemble cast. This is a further, shared step in the devaluation of reason, one that does not depend on the excesses of Treatise I.iv.1, I.iv.4, and I.iv.7.22

22. I thank Victor Caston, Gary Hatfield, James Joyce, Howard Nye, Frederick Schmitt, Matthew Silverstein, and Lawrence Sklar for helpful comments and discussion. I am indebted to Alex Silk for providing detailed substantive, organizational, and stylistic comments.

introduction

33

1 Is There Radical Dissimulation in Descartes’ Meditations?

1. Introduction according to dissimulation hypotheses, Descartes, in the Meditations, intentionally misrepresented important aspects of his philosophy. There are a number of versions of such dissimulation hypotheses, depending on where the points of misrepresentation are located.1 For the purposes of this essay, I am interested in the thesis that Descartes was not sincere either about his proofs of the existence of God in Meditation III or about his appeal to Divine veracity in order to validate clear and distinct perception in Meditation IV and in order to prove the existence of the material world in Meditation VI.2 I will call this thesis the dissimulation hypothesis. I believe there is a serious possibility that the dissimulation hypothesis is correct and hence that a number of claims at the core of the Meditations do not represent Descartes’ considered philosophical views. My purpose in this essay is to explore this possibility. I will sketch quite briefly two considerations that are strongly suggestive of dissimulation with respect to the epistemological role assigned to God in the Meditations. The first consideration relates to the problem of the Cartesian circle. The texts that serve as the initial stimulus for dissimulation hypotheses are precisely those which evidence circularity in the argument of 1. There are two contemporary proponents of the view that Descartes is engaged in extensive dissimulation in the Meditations. See Caton 1970, 1971, 1973, and 1975, esp. 97–100, and Dorter 1973. For earlier versions of dissimulation hypotheses, see Adam 1910, 304–7, and Leroy 1929, esp. 1.15–21, 2.17–42. 2. This thesis is not meant to suggest that Descartes was not pious, or not a theist, or not sincere about his proof of the existence of God in Meditation V.

34

Meditations III and IV.3 Commentators have found the procedure of Meditations III and IV question-begging on the ground that Descartes must prove the existence of a nondeceiving God in order to validate clear and distinct perception; however, he must rely on clear and distinct perception in conducting these proofs. That there is the strong appearance of circularity is not controversial. In the Fourth Objections, Arnauld accused Descartes of “circular reasoning” (HR 2.92: AT 7.214).4 Gassendi objected repeatedly that there was a circular argument beginning in Meditation III.5 A closely related objection is raised in the Second Objections (HR 2.26: AT 7.124–25). The circularity has seemed so obvious that one wonders whether Descartes could have failed to be aware of it. The second consideration relates to the substance of the arguments for the existence of God in Meditation III.6 The two arguments are notoriously weak.7 Both rely on the principle that there must be at least as much formal (actual) perfection in the efficient and total cause of an idea as objective perfection in the idea itself. Unfortunately, it is difficult to see what there is to recommend this principle other than its suitability for Descartes’ argumentative purposes.8 One wonders how a figure of Descartes’ intelligence, who has undertaken “to withhold . . . assent from matters which are not entirely certain and indubitable” (HR 1.145: AT 7.18), and who includes beliefs about mathematics within the scope of the doubt (HR 2.158–59: AT 7.35–36), could nevertheless unhesitatingly embrace the principle about causation as a deliverance of the light of nature.9 In sum, Descartes’ argument in Meditations III and IV seems so obviously question-begging as to suggest the possibility that Descartes was himself 3. Curley treats the Cartesian circle as the principal indication of dissimulation, though Curley himself rejects the dissimulation hypothesis. See Curley 1978, 96–98. The problem of circularity plays a significant role in Dorter’s defense of the dissimulation hypothesis. See Dorter 1973, 318–20, 326. 4. [“Abbreviations for Editions of Seventeenth- to Nineteenth-Century Works” contains the full reference system for this chapter. Section numbers have been added in the text.] 5. See Pierre Gassendi, Metaphysical Colloquy, or Doubts and Rebuttals concerning the Metaphysics of René Descartes, with his Replies, Rebuttals to Med. III, Doubt I, Article 1; Rebuttals to Med. IV, Doubt I, Article 1, and Doubt IV, Article 2. The relevant passages are translated at Br 204, 231, 241–42. 6. Dorter also finds characteristics of the arguments for the existence of God in Meditation III suggestive of dissimulation (1973, esp. 326–27). 7. See, for example, Kemp Smith 1952, 302–3; Kenny 1968, 20, 126–45; and B. Williams 1978, 142–52. 8. For criticism of the principle about causation, see A. B. Gibson 1932, 110–24; B. Williams 1978, 138–41, 142–43; and M. Wilson 1978, 136–38. 9. B. Williams writes of Descartes’ acceptance of the principle that there must be at least as much perfection in the efficient and total cause as in its effect: “It is the question of the causation of his ideas that Descartes now pursues. And here he makes a sudden jump forward, receiving a deliverance from the ‘natural light’ at once more substantial and less plausible than many propositions about which he has felt qualms at earlier stages of his reflection. . . . This is a piece of scholastic metaphysics, and it is one of the most striking indications of the historical gap that exists between Descartes’ thought and our own . . . that he can unblinkingly accept this unintuitive and barely comprehensible principle as self-evident in the light of reason” (1978, 135).

dissimulation in descartes’ MEDITATIONS

35

aware of the circularity, and hence not sincere in his appeal to the existence of a nondeceiving God in order to validate clear and distinct perception. The considerations relating to the problem of the Cartesian circle are reinforced by those relating to the substance of the arguments for the existence of God in Meditation III. Such glaring deficiencies as the appeal to the intrinsically implausible causal principle suggest the possibility that Descartes was not sincere in proposing them. This is one way to sketch a prima facie case for dissimulation, with respect to both the arguments for the existence of God in Meditation III and the appeal to Divine veracity in order to validate clear and distinct perception in Meditation IV.10 While I believe that this prima facie case can be considerably strengthened, my goal in this essay is neither to elaborate on the considerations I have sketched nor even to argue directly for the truth of the dissimulation hypothesis. My objective is to remove certain objections to the acceptance of a dissimulation hypothesis of the sort I have introduced. Two obstacles are frequently discussed. The first is the methodological problem of developing general criteria for interpretation under the assumption of dissimulation.11 The second is a reluctance to attribute to Descartes the defects of character—dishonorableness, cowardice, and so forth—which the dissimulation hypothesis might seem to entail.12 The obstacles I have in mind are more purely interpretive and philosophical. The arguments for the existence of a nondeceiving God are prompted by the doubt “concerning things which seemed to me most manifest” (HR 1.158: AT 7.36) raised at the fourth paragraph of Meditation III. The obstacle is simply this: what are we to make of this doubt on the hypothesis that Descartes was not sincere in his response to it, was not sincere in his claims about God’s epistemological role in Meditations III and IV? In §2, I argue that there is textual evidence that Descartes in fact minimizes the importance of the doubt of Meditation III. In §3, I locate in Descartes a rationale for not taking the doubt of Meditation III seriously. The outcome is to identify in Descartes a nontheological epistemological position compatible with the dissimulation hypothesis. In §4, I provide a brief account of the motives Descartes possessed for engaging in dissimulation.

10. The prima facie case does not directly support one component of the dissimulation hypothesis. Suppose that Descartes was sincere in proposing the ontological argument for the existence of God in Meditation V. The use of the ontological argument does not beg the question against the doubt about the existence of the material world in Meditation I. It is thus consistent with the considerations in the text that Descartes was sincere in his appeal to Divine veracity in order to prove the existence of the material world. This suggests a weakened version of the dissimulation hypothesis, according to which the dissimulation does not extend to the appeal to Divine veracity in Meditation VI. I have more to say about the weakened version of the dissimulation hypothesis in notes 29 and 40. 11. See Laporte 1950, 299–300, 465–66, and Caton 1970, 12–15. 12. See Caton 1973, 15–20, and Dorter 1973, 339.

36

reflection and the stability of belief

2. The Slight and Metaphysical Doubt of Meditation III In the second paragraph of Meditation III, Descartes states: “it seems to me that already I can establish as a general rule” (HR 1.158, emphasis added: AT 7.35) that whatever is clearly and distinctly perceived is true. It would, however, be premature to take this general rule as established. In the fourth paragraph, Descartes introduces the following ground for doubt: But when I took anything very simple and easy in the sphere of arithmetic or geometry into consideration, e.g. that two and three together made five, and other things of the sort, were not these present to my mind so clearly as to enable me to affirm that they were true? Certainly if I judged that since such matters could be doubted, this would not have been so for any other reason than that it came into my mind that perhaps a God might have endowed me with such a nature that I may have been deceived even concerning things which seemed to me most manifest. But every time that this preconceived opinion of the sovereign power of a God presents itself to my thought, I am constrained to confess that it is easy to Him, if He wishes it, to cause me to err, even in matters in which I believe myself to have the best evidence. (HR 1.158: AT 7.35–36) In the context, it is natural to take the doubt about “things which seemed to me most manifest,” or “matters in which I believe myself to have the best evidence,” as relating to clear and distinct perception.13 No comparable doubt is introduced in the Rules for the Direction of the Mind. This is an early, and unfinished, work; however, it contains Descartes’ most extended account of doctrines relating to clear and distinct perception. In the Rules, Descartes employs the technical term ‘intuition’ (‘mental vision’) to refer to an act or operation of the mind in which a proposition is perceived all at once or in a moment, and so clearly and distinctly as to be certain or indubitable (cf. Rules III, VII, XI). Deduction does not rely on any second basic cognitive faculty distinct from intuition; a deduction consists of a sequence of connected intuitions. In the Rules, Descartes writes that “deduction, or the pure illation of one thing from another . . . cannot be erroneous when performed by an understanding that is in the least degree rational” (HR 1.4–5, emphasis added: AT 10.365). As for intuition, “it is more certain than deduction itself, in that it 13. Descartes states in the fourth paragraph of Meditation III that the doubt he raises there can only be removed by proving the existence of a nondeceiving God. In the final paragraph of Meditation IV, Descartes appeals to the existence of a nondeceiving God in order to sustain the conclusion that whatever is clearly and distinctly perceived is true. It is therefore difficult to resist the conclusion that the doubt raised in the fourth paragraph of Meditation III is about clear and distinct perception.

dissimulation in descartes’ MEDITATIONS

37

is simpler, though deduction, as we have noted above, cannot by us be erroneously conducted” (HR 1.7, emphasis added: AT 10.368). Descartes thus introduces the discussion of intuition and deduction at Rule III as follows: “we shall here take note of all those mental operations by which we are able, wholly without fear of illusion, to arrive at the knowledge of things” (HR 1.7, emphasis added: AT 10.368). In the Rules, there is no hint of skepticism about intuition, the mental act involving clear and distinct perception. Descartes treats intuition as unproblematic and exhibits nothing but confidence in it. Against this background, the introduction of the skeptical hypothesis of Meditation III constitutes a surprising about-face.14 The question that arises is how to explain the discrepancy between Descartes’ treatment of clear and distinct perception in the Rules and in the Meditations. There are at least three possible lines of interpretation. The first is that the discrepancy reflects a significant change in Descartes’ view about the epistemological status of intuition or clear and distinct perception. Both the second and third interpretations deny this, holding that any change in view is more apparent than real; these interpretations differ as to how the (merely) apparent change in view is to be explained. On the second interpretation, the Rules and the Meditations simply address different problems; the Rules is concerned with the method proper, whereas the Meditations is concerned with the metaphysical foundations of the method. On the third interpretation, Descartes was not sincere in raising the doubt about clear and distinct perception in Meditation III. It is this third interpretation that is congenial to the dissimulation hypothesis. It is easier to believe that Descartes was not sincere in his claims for God’s epistemological role if he was not sincere about the doubt that prompts the proofs of the existence of a nondeceiving God. There are a number of features of the way in which Descartes raises the doubt in the fourth paragraph of Meditation III which suggest that he may not have taken the doubt seriously. These features emerge most clearly if one compares the hypotheses that generate radical doubt in Meditations I and III, respectively. The doubts raised in these two Meditations differ in scope, that is, in the classes of beliefs to which they apply. In Meditation I, Descartes introduces the hypothesis of an omnipotent deceiver who causes him to have sensory experiences as if material objects existed; this hypothesis generates a doubt about beliefs based on sensory experience.15 In Meditation III, Descartes introduces the hypothesis of an omnipotent deceiver who causes him to have false beliefs about matters that seem most certain; this hypothesis generates a doubt about beliefs based on clear and distinct perception or intuitive apprehension. 14. The discrepancy has been noted, and variously explained. See Beck 1952, 38–43, and 1965, 134–35; Caton 1973, 48; Curley 1978, viii, 35–38, 103; and Keeling 1968, 15–16, 79–81, 84–85. 15. I follow Frankfurt in taking the doubt of Meditation I as extending to beliefs about mathematics only insofar as they are taken to be based on the senses. See Frankfurt 1970, pt. 1, esp. chs. 7–8.

38

reflection and the stability of belief

Whereas Descartes describes the grounds for the doubt of Meditation I as “very powerful and maturely considered” (HR 1.148: AT 7.21), he describes the grounds for the doubt of Meditation III as “very slight, and so to speak metaphysical” (HR 1.159: AT 7.36).16 This marked contrast in tone cries out for explanation.17 The two hypotheses of a deceiver seem relevantly parallel. Both hypotheses are offered as representing epistemic possibilities: for all Descartes knows, at the points in the Meditations where the respective hypotheses are advanced, it is possible that they are true. For all Descartes knows, at the close of Meditation I, there might exist an omnipotent being who causes him to have false beliefs based on sensory experience; for all Descartes knows, early in Meditation III, there might exist an omnipotent being who causes him to have false beliefs based on clear and distinct perception. In neither case does Descartes make any claim about the likelihood or probability that the hypothesis is true.18 In these circumstances, it is difficult to find a rationale for the description of one doubt as “very slight, and so to speak metaphysical” and the other doubt as “very powerful and maturely considered.” Descartes states that the ground for doubt in Meditation III is slight and metaphysical “since I have no reason to believe that there is a God who is a deceiver, and as I have not yet satisfied myself that there is a God at all” (HR 1.159: AT 7.36). But it is equally true, in the context of Meditation I, that he has no reason to believe in the existence of a deceiving God and that he has not yet proved that God exists. The two hypotheses seem to share the same epistemological status, and hence the doubts they generate should be equally slight, or equally powerful.19 Why, 16. Some commentators illicitly import the description of the ground for doubt in Meditation III into Meditation I. Beck writes: “But in the first Meditation even the truths of arithmetic are shown to be open to doubt. The doubt itself may be, as Descartes says, ‘very slight’: it may only be a ‘metaphysical doubt which leads me to question that 2 +2 = 4’” (1952, 40). Beck’s citation (40) reads “cf. Meditationes, I (AT vii. 3624–26).” The AT passage, of course, is found in Meditation III, not Meditation I. Also see Buchdahl 1969, 157–58. 17. The issue is almost never raised explicitly. Curley is one commentator who sees that there is an interpretive problem: “the ground of doubt which Descartes here characterizes as ‘valid and carefully considered’ he will later . . . call ‘slight and metaphysical.’ This is puzzling and will require explanation” (1978, 42–43). Prichard writes: “Though Descartes says the ground of doubt is slight, it is not really so. The ground is serious. . . . He adds that it is metaphysical. . . . To say this does not make the difficulty any less serious” (1950, 85). Prichard fails to ask the obvious question: why does Descartes say that the doubt of Meditation III is slight? 18. In Meditation I, Descartes does state: “I have long had fixed in my mind the belief that an all-powerful God existed” (HR 1.147: AT 7.21). The belief in an omnipotent God is a preconceived opinion. However, the belief in an omnipotent God who is a deceiver is not a preconceived opinion. And even if it were, this fact would not account for the difference in Descartes’ description of the grounds for doubt in Meditations I and III. This is because the belief in an omnipotent God (whether or not such a being is believed to be a deceiver) would have the same status as a preconceived opinion both at the end of Meditation I and at the beginning of Meditation III, prior to the arguments for the existence of God. 19. The literature is of little help here. M. Wilson simply repeats Descartes’ own inadequate explanation: “Since [Descartes] has no reason to believe there is such a [deceiving] God, the reason for doubt that depends on this idea is ‘very tenuous and so to speak Metaphysical’” (1978, 120; cf. 130). Kenny writes: “Descartes sometimes calls his doubt ‘hyperbolical’ and ‘metaphysical’

dissimulation in descartes’ MEDITATIONS

39

then, does Descartes employ language that minimizes the doubt introduced in Meditation III as compared to that introduced in Meditation I? There are two respects in which the hypothesis of Meditation III is disanalogous to the hypothesis of Meditation I.20 Consider the latter. How could it be the case that one’s belief in the existence of a material world, where that belief is based on the sensory evidence in fact available to one, is false? Descartes is quite explicit: perhaps God “has . . . brought it to pass that there is . . . no extended body, and that nevertheless [I possess the perceptions of all these things and that] they seem to me to exist just exactly as I now see them” (HR1.147: AT 7.21). The hypothesis has two components: first, an explanation of how it could be the case that no material objects exist (an omnipotent being has caused it to be the case that none exists); and second, an explanation of how one could nevertheless have sensory experiences as if material objects existed (an omnipotent being has caused one to have those sensory experiences). Descartes’ hypothesis explains how the belief in a material world could be false, and how one could have the evidence one does have for that belief in the circumstances that it is false. The hypothesis of Meditation III seems entirely to lack any analogue to the first component of the hypothesis of Meditation I. The belief in the existence of a material world would be false if an omnipotent being “brought it to pass that there is . . . no extended body.” In Meditation III, Descartes offers no explanation as to how one’s most certain beliefs, such that two and three together make five, could be false. It is tempting to suppose that we

(AT VII, 37, 90; HR I, 159, 199). By this he means that the suppositions on which the doubt depends—that life is a dream, that there is an omnipotent deceiver—are very improbable suppositions” (1968, 23–24). Prior to the proofs of the existence of a nondeceiving God, Descartes would seem to have no resources to show that the hypotheses in question are improbable (except in a purely subjective sense). But the important point is that the suppositions of an omnipotent deceiver in Meditations I and III would seem to be equiprobable, and hence Kenny cannot explain why the reasons for the former are “very powerful.” Although Curley undertakes to explain the difference in the characterizations of the grounds for doubt in Meditations I and III (see note 17), I do not find a satisfactory explanation in his discussion. Curley emphasizes that the hypothesis of an omnipotent deceiver is a reasonable ground for doubt prior to the end of Meditation IV (cf. 1978, 106–7, 116, 119–20). It follows, as Curley admits (cf. 124) that it is also a reasonable ground for doubt at the beginning of Meditation III. 20. There is a difference I do not discuss. In Meditation I, Descartes initially formulates the hypothesis with reference to an omnipotent God (¶9) and subsequently formulates the hypothesis of the evil genius (¶12). In Meditation III, the hypothesis is formulated exclusively with reference to God. In advance of a proof of the existence of a nondeceiving God, this difference seems irrelevant to the force of the respective hypotheses. In each Meditation, the hypothesis in question is that there exists a powerful being who deceives Descartes in particular respects. What is more, the texts preclude assigning any epistemological significance to the difference in question. In Meditation I, the ground for doubt is described as powerful and mature on the basis of the initial formulation with reference to God, prior to the formulation with reference to a genius. Both contrasting characterizations of the grounds for doubt thus refer to hypotheses about Divine deception.

40

reflection and the stability of belief

have located an explanation of why Descartes treats the doubt raised in Meditation III as slight and metaphysical in contrast to the powerful and mature reasons for the doubt raised in Meditation I. The hypothesis of Meditation I, unlike that of Meditation III, includes an explanation of how the beliefs in question could be false. Descartes describes circumstances in which the belief in a material world would be false; he does not describe circumstances in which his most certain beliefs would be false. The doubt of Meditation III then appears slight in the sense that Descartes at best asserts that it is possible that it is false that, for example, two plus three equals five, without describing how this is possible. Unfortunately, this attempt to explain why Descartes treats the doubt of Meditation III as slight is not satisfactory. It is true that in Meditation III Descartes does not provide an explanation of how it could be false that two plus three make five. However, he could have provided such an explanation by invoking his body of doctrine about the “eternal truths” (see K 11, 13–15, 150–51, 236–37; HR 2.226, 248; C 33: AT 1.145–46, 149–53, 4.118–19, 5.223–24, 7.380, 431–32, 5.159–60). The eternal truths approximately coincide in content with what might be called presumptively necessary truths. According to Descartes’ doctrine, the eternal truths are in fact contingent or dependent upon God’s will. Descartes’ ground for this is simply that God would not be omnipotent were He bound by any truths that were strictly necessary in the sense of being independent of His will. Further, Descartes holds that God established the eternal truths “by the same kind of causality as he created all things” (K 14: AT 1.151–52). Descartes states his doctrine of eternal truths as early as 1630, as late as 1648, and in his Fifth and Sixth Replies. In light of this body of doctrine, Descartes could have hypothesized in Meditation III that an omnipotent deceiver “brought it to pass” that it is false that two plus three make five, much as in Meditation I he hypothesized that an omnipotent deceiver “brought it to pass” that there exists no extended body. The doctrine of the eternal truths, however, is not mentioned in the Meditations.21 Thus, we have not succeeded in explaining why Descartes treats the doubt of Meditation III as slight. An interesting question is why Descartes does not avail himself of this doctrine in order to establish a greater parity between the hypotheses of Meditations I and III.22 Suppose Descartes did take the doubt of Meditation III as seriously as that of Meditation I and that he was committed to his doctrine about the eternal truths. Under these circumstances, we would expect him to invoke this doctrine in order to provide an explanation of how our beliefs in presumptively necessary truths could be false.23 But 21. Numerous commentators agree on this point. See, for example, Beck 1952, 8; Bréhier 1937, 193; Caton 1973, 68; and Frankfurt 1970, 7, and 1977, 37. 22. For two commentators who maintain that the doubt of Meditation III does depend upon the doctrine of the eternal truths, see Bréhier 1937 and M. Wilson 1978, esp. 33–34, 121, 128. 23. It is frequently maintained that Descartes had prudential reasons for withholding his doctrine about the creation of the eternal truths. Cf. Bréhier 1937, 193; Frankfurt 1970, 7; and A. B. Gibson 1932, 53. This might be taken to defeat the expectation that Descartes would

dissimulation in descartes’ MEDITATIONS

41

Descartes does not do so. This suggests that either Descartes did not take the doubt of Meditation III seriously, that he did not take his doctrine about the eternal truths seriously, or both. It should be noted that the position that he was serious about the doubt of Meditation III, but not about the doctrine of the eternal truths, concedes that outside of the Meditations Descartes was involved in dissimulation with respect to his views about Divine omnipotence. Consider the second component of the skeptical hypothesis of Meditation I. This component provides an explanation of how one could have sensory experiences as if material objects existed in the circumstances that they do not exist. On the hypothesis of Meditation I, God “brought it to pass that there is no earth, no heaven, no extended body, no magnitude, no place, and that nevertheless [I possess the perceptions of all these things and that] they seem to me to exist just exactly as I now see them” (HR 1.147: AT 7.21). How does God bring it to pass that one has such sensory experiences? Perhaps God endows the mind with a faculty or capacity which itself generates the sensory experiences. It is clear from the passage in Meditation VI where Descartes reconsiders the hypothesis of Meditation I that this is not the mechanism he envisions. Descartes writes: “since God is no deceiver, it is very manifest that He does not communicate to me these ideas [of sensible things] immediately and by Himself, nor yet by the intervention of some creature in which their reality is not formally, but only eminently, contained” (HR 1.191: AT 7.79). In other words, the hypothesis of Meditation I involves an omnipotent being who causes one’s sensory experiences directly in the sense that he induces them one by one.24 In offering this particular explanation, Descartes commits himself to a definite argumentative strategy. To see what I have in mind, let us distinguish between beliefs about the material world and the mechanisms that might cause those beliefs. For example, beliefs about the material world might be based on sensory experiences that are caused by mechanisms involving causal interactions between extended bodies and the sense organs and the brain, that is, by mechanisms of the sort which we take to be operative in connection with everyday beliefs about the material world. I will refer to such mechanisms as sense-perception. Alternatively, beliefs about the material

invoke the doctrine for the suggested purpose in the Meditations. I believe Descartes was sincere in his doctrine about the eternal truths and that he did suppress the doctrine for reasons of prudence. I also believe, however, that since he had reasons for not taking the doubt of Meditation III seriously (see §3), he would not have wanted to appeal to this doctrine in order to maintain parity between the hypotheses of Meditations I and III. 24. At the close of the ninth paragraph of Meditation I, Descartes writes of the possibility that God might have created me such that “I am always mistaken” (AT 7.21: HR 1.147). This might be taken to suggest that in Descartes’ hypothesis God creates one with a faculty such that one is always mistaken. The passage is at best suggestive of this interpretation, however, and does not require it. I think it reasonable to resolve the ambiguity in favor of the explicit passage in Meditation VI.

42

reflection and the stability of belief

world might be based on sensory experiences that are caused directly by the volitions of an immaterial omnipotent deceiver. Two strategies are available to challenge beliefs about the material world. The first is to suggest that sense-perception might be intrinsically defective in the sense that it is an unreliable belief-forming mechanism (or set of belief-forming mechanisms). (An unreliable belief-forming mechanism is one that leads to false beliefs more often than not—either in actual or in relevant possible circumstances.) The second strategy is to suggest that beliefs about the material world might be caused by some unreliable mechanisms distinct from sense-perception, such as the direct volitional activity of an omnipotent deceiver. The second strategy leaves open the question of whether sense-perception itself is reliable. It is this second strategy that Descartes employs. Descartes never uses the first strategy; he never raises the possibility that sense-perception is intrinsically defective. He merely notes that “it is sometimes proved to me that these senses are deceptive, and it is wiser not to trust entirely to any thing by which we have once been deceived” (HR 1.145: AT 7.18). Here, Descartes calls attention to the fallibility of sense-perception, making the point that beliefs based on sense-perception are sometimes false. He does not question the reliability of sense-perception by suggesting that beliefs based on sense-perception might be false more often than not.25 Descartes abstains from directly challenging the reliability of his perceptual faculties, that is, of the faculty of sense-perception. I believe Descartes considers the possibility of such a challenge and consciously declines to employ it. This is the significance, at least in part, of the cryptic discussion of madness in the fourth paragraph of Meditation I. Descartes asks: And how could I deny that these hands and this body are mine, were it not perhaps that I compare myself to certain persons, devoid of sense, whose cerebella are so troubled and clouded by the violent vapours of black bile, that they constantly assure us that they think they are kings when they are really quite poor, or that they are clothed in purple when they are really without covering, or who imagine that they have an earthenware head or are nothing but pumpkins or are made of glass. (HR 1.145: AT 7.18–19) Two of the three examples are beliefs about perceptible features of the world. This suggests that Descartes is raising the possibility that he has unreliable perceptual faculties.26 His response is to dismiss the hypothesis: such persons “are mad, and I should not be any the less insane were I to follow examples so extravagant” (HR 1.145: AT 7.19). Descartes is not rejecting 25. Cf. Frankfurt 1970, 34. 26. For a different interpretation of the madness passage, see Frankfurt 1970, 37–38. The interpretation of Caton 1973, 111, is congenial to my own.

dissimulation in descartes’ MEDITATIONS

43

the hypothesis of defective perceptual faculties as an epistemic possibility. For all he knows, the hypothesis could be true. He is declining to invoke the possibility that sense-perception might itself be defective as a ground for doubt about beliefs based on sensory experience. Thus, in Meditation I, the possibility of defective or unreliable perceptual faculties is raised, and dismissed, in paragraph four. Descartes chooses to use the second strategy I have described, hypothesizing that sensory experiences are caused directly by an omnipotent deceiver. Consider the second component of the hypothesis of Meditation III. It falls to this component to explain how one could have false beliefs in presumptively necessary truths. It is the role of the omnipotent being “to cause me to err, even in matters in which I believe myself to have the best evidence.” This hypothesis has a superficial parity to that of an omnipotent being who causes me to have false beliefs about the material world. The apparent parity vanishes, however, if one asks how the omnipotent being of Meditation III causes false beliefs. There are two possibilities. The first is that the omnipotent being causes one to have a cognitive faculty—“intuition,” “reason,” “the light of nature,” whatever—which is intrinsically defective, unreliable in the sense that its normal operation leads one to misapprehend falsehoods as truths. This would be analogous to the hypothesis that our perceptual faculties are defective. The second possibility is that the omnipotent being directly causes one to have the clear and distinct perceptions or intuitive apprehensions on which the false beliefs are based. This would be analogous to the mechanism embodied in the hypothesis of Meditation I, where the omnipotent being directly causes our sensory experiences. The hypothesis of Meditation III seems to rely on the first mechanism, on which the false beliefs issue from a defective cognitive faculty. Descartes writes: “it came into my mind that perhaps a God might have endowed me with such a nature that I may have been deceived even concerning things which seemed to me most manifest” (HR 1.158, emphasis added: AT 7.36). This strongly suggests that Descartes is envisioning a being who endows him with some stable faculty, the operation of which is the proximate cause of his misapprehending falsehoods as truths. Suppose Descartes was envisioning a being who directly causes his misperception or misapprehension of falsehoods as truths. If so, it would be misleading to write of the being’s endowing him with a nature such that he is deceived, for by hypothesis there would be nothing in Descartes’ nature, nothing about his faculties, to account for the false beliefs. The development of the hypothesis of Meditation III, unlike that of Meditation I, depends upon the possibility that a cognitive faculty is unreliable.27

27. Frankfurt maintains that the hypothesis of Meditation I does raise “the possibility that the human mind is inherently defective” (1970, 81). I have argued previously that this is not the most plausible interpretation of the hypothesis of the omnipotent deceiver. As far as I can see, Frankfurt does not so much argue for his interpretation as presuppose it; he does not directly

44

reflection and the stability of belief

Once again, it is tempting to suppose that we have located an explanation of why Descartes treats the doubt of Meditation III as slight in contrast to the powerful doubt of Meditation I. Descartes’ explanation of how one could have false beliefs about the material world does not rely on the supposition that one’s perceptual faculties themselves are unreliable; Descartes’ explanation of how one could have false beliefs about necessary truths does rely on the supposition that one’s rational faculties are unreliable.28 Descartes could have maintained parity, however, between the second components of the hypotheses of Meditations I and III; he could have supposed that the omnipotent being causes false mathematical beliefs by causing clear and distinct perceptions directly, rather than by endowing us with a defective faculty of reason. The situation is similar to that in connection with the first component of the two hypotheses. Descartes explains how our beliefs about the material world could be false; he does not explain how our beliefs about necessary truths could be false. He could have maintained parity, however, between the first components of the hypotheses of Meditations I and III; he could have appealed to his doctrine of the eternal truths in order to explain how our beliefs about putatively necessary truths could be false. There are thus two disanalogies between the omnipotent deceiver hypotheses of Meditations I and III as these hypotheses are presented by Descartes. However, Descartes could have constructed the two hypotheses in such a way as to remove these disanalogies. Apparently, Descartes preferred to build disanalogies into the two hypotheses, thereby enabling him to minimize the doubt of Meditation III. In §1, I presented a prima facie case for the following claims: that at the time he wrote the Meditations, Descartes was aware of the circularity in his argument in Meditations III and IV, and that he did not himself accept the arguments for the existence of God in Meditation III. If so, Descartes was not sincere about the epistemological role assigned to God in Meditations III and IV. This seems implausible if Descartes was sincere about the doubt of Meditation III which prompts these developments. I have suggested that Descartes constructs the hypothesis of Meditation III in a way that enables him to minimize the doubt it generates, that is, to treat the doubt as slight in

consider my alternative to it. The term ‘nature’ does not occur in the context of the development of the hypothesis of Meditation I. It is sufficient for my purposes that the hypothesis of Meditation III, unlike the hypothesis of Meditation I, is stated in such a way as to require that the mind has an intrinsically defective faculty. 28. Why does this render the doubt of Meditation I comparatively strong? The intuitive idea is this. Suppose a skeptic generates doubt with respect to beliefs about the material world by suggesting that the faculty of sense-perception might itself be defective. Although this hypothesis does not seem illegitimate, it does seem to beg the question directly against the nonskeptic. The suggestion that our perceptual beliefs are caused by some unreliable mechanism distinct from sense-perception does not beg the question about the reliability of sense-perception itself. In this sense, the latter ground for doubt is stronger. The same points would apply to doubts relating to reason.

dissimulation in descartes’ MEDITATIONS

45

contrast to the doubt of Meditation I. If this is a signal that Descartes did not take the doubt of Meditation III seriously at all, then we can view his introduction of that doubt as nothing but a pretext for the subsequent arguments in Meditations III and IV. This raises a question as to how Descartes could have failed to take the doubt of Meditation III seriously. We have to locate in Descartes an epistemological position that enables us to answer this question.

3. Cartesian Epistemology without Divine Veracity One way to address this question is by considering an obvious objection to dissimulation hypotheses. In its most general form, the difficulty is this: if Descartes was not sincere in his appeal to Divine veracity in order to validate our cognitive faculties, what was his epistemological position? The following considerations generate a particularly acute form of the difficulty. The doubt raised in Meditation III extends to those “matters in which I believe myself to have the best evidence,” that is, to those beliefs which are presumed to be certain even if no material objects exist. According to the dissimulation hypothesis, this doubt is a pretext for locating an apparent epistemological role for God. In Meditation I, however, Descartes has raised a distinct doubt about the existence of material objects. In Meditation VI, Descartes argues that since he has “a very great inclination to believe . . . that [ideas] are conveyed to him by corporeal objects” (HR 1.191: AT 7.79–80), God would be a deceiver if material objects did not exist. If Descartes was not sincere in his appeals to Divine veracity, the argument for the existence of a material world collapses, leaving Descartes with no response to the doubt raised in Meditation I.29 Proponents of the dissimulation hypothesis have failed to respond adequately to this objection. Suppose that the argument for the existence of the material world is itself dismissed as not constituting serious Cartesian doctrine. In these circumstances, a dissimulation theorist must embrace one of three alternative positions. The first alternative is that Descartes was not sincere even about the doubt raised in Meditation I.30 This variant of the dissimulation hypothesis cannot explain why the doubt of Meditation I is treated as based on considerations that are powerful and mature. The second alternative is that Descartes was serious about the doubt of Meditation I but failed to supply any sincere response to the doubt. This variant of the dissimulation hypothesis has the consequence that the epistemological

29. This “acute form of the difficulty” does not arise for the weakened version of the dissimulation hypothesis (see note 10). 30. As Curley notes, “Dissimulation theorists typically do not see much force in the skeptical arguments Descartes professes to reply to” (1978, 100).

46

reflection and the stability of belief

problem of Meditation I is left unresolved in the Meditations. The third alternative is that Descartes was sincere about the doubt of Meditation I and does provide a response to that doubt which does not rely on Divine veracity. I believe that, if a dissimulation hypothesis is to receive a hearing, it will have to be coupled with this third alternative. In this third part of my essay, I sketch one version of an interpretation along the lines of the third alternative. The project is to locate a Cartesian response to the doubt of Meditation I, and more generally an overall Cartesian epistemological position, which does not rely on the appeals to Divine veracity contained in the Meditations. I believe that such a theory can be extracted from Descartes’ writings. Central to the nontheological epistemological position which I locate in Descartes is the conception of a hierarchy of cognitive faculties. This has important analogies to the conception of a hierarchy, pyramid, or vertical structure of beliefs in the context of foundation theories of justification. The foundationalist defines an asymmetric relation of epistemological priority among beliefs: the belief that p is epistemologically prior to the belief that q just in case one’s belief that p can be justified without appeal to the belief that q, whereas one’s belief that q cannot be justified without appeal to the belief that p; the belief that p is epistemologically basic just in case there is no belief epistemologically prior to it. For example, in standard interpretations of Descartes, the belief “I exist” (or certain closely related beliefs) is epistemologically basic, and epistemologically prior to the belief “God exists,” which is in turn epistemologically prior to the belief “there exists a material world.” I believe that any such theory in Descartes is derivative from a more fundamental foundational theory of cognitive faculties. In order to generate the relevant hierarchical structure, we must define an asymmetric relation of epistemological priority among cognitive faculties. There is an instructive passage in the Rules: This furnishes us with an evident explanation of the great superiority in certitude of Arithmetic and Geometry to other sciences. The former alone deal with an object so pure and uncomplicated, that they need make no assumptions at all which experience renders uncertain, but wholly consist in the rational deduction of consequences. (HR 1.5: AT 10.364) Descartes does not simply assert that mathematical beliefs based on reason are superior in certainty to beliefs based on sense-perception. He sketches an explanation: reason, when applied to mathematics, requires no assumptions that experience or sense-perception renders uncertain. In other words, sense-perception provides no grounds for uncertainty about mathematical beliefs based on reason. The point is generalized in Meditation III. Descartes writes of the natural light, which includes intuition or clear and distinct perception: “And I possess no other faculty whereby to distinguish truth from falsehood, which can teach me that what this light shows me to

dissimulation in descartes’ MEDITATIONS

47

be true is not really true” (HR 1.160–61: AT 7.38–39).31 In other words, there is no cognitive faculty that can show that any belief based on reason is false. In the passage about the natural light, Descartes adds that he possesses “no other faculty that is equally trustworthy” (HR 1.161: AT 7.38). In the context, reason appears trustworthy precisely because beliefs based on reason cannot be shown to be false by any other cognitive faculty. Descartes’ discussion suggests that beliefs based on any other cognitive faculty can be shown to be false by reason. This seems to be confirmed two paragraphs later with respect to the central case of sense-perception: I find, for example, two completely diverse ideas of the sun in my mind; the one derives its origin from the senses and, should be placed in the category of adventitious ideas; according to this idea the sun seems to be extremely small; but the other is derived from astronomical reasonings, i.e. is elicited from certain notions that are innate in me, or else it is formed by me in some other manner; in accordance with it the sun appears to be several times greater than the earth. These two ideas cannot, indeed, both resemble the same sun, and reason makes me believe that the one which seems to have originated directly from the sun itself, is the one which is most dissimilar to it. (HR 1.161: AT 7.39) Beliefs based on sense-perception about the size of astronomical bodies can be shown to be false by reason.32 In sum, beliefs based on sense-perception can be shown to be false by reason, but beliefs based on reason cannot be shown to be false by sense-perception, or by any other cognitive faculty. Descartes’ account of the relationship between sense-perception and reason suggests a general definition of an asymmetric relation of epistemological priority among cognitive faculties. Let us say that if a belief based on cognitive faculty f has been shown to be false by cognitive faculty g, then the belief based on f has been corrected by cognitive faculty g. Cognitive faculty f is epistemologically prior to cognitive faculty g just in case beliefs based on g can be corrected by f, whereas beliefs based upon f cannot be corrected by

31. “[I]ntuition is the undoubting conception of an unclouded and attentive mind, and springs from the light of reason alone” (HR 1.7: AT 10.368); “the light of nature, or the faculty of knowledge which God has given us, can never disclose to us any object which is not true, inasmuch as it . . . apprehends it clearly and distinctly” (HR 1.231: AT 8A.16). 32. There is another example. Descartes maintains that the mistaken judgment, based on sight, that a stick protruding from water is bent, is corrected by reason (HR 2.252–53: AT 7.438–39). This is true even if the stick feels straight to the touch: “we need to have some reason to show why in this matter we ought to believe the tactual judgment rather than that derived from vision; and this reason . . . must be attributed not to sense but to the understanding. Hence in this instance it is the understanding solely which corrects the error of sense” (HR 2.253: AT 7.439). Descartes concludes that “no case can ever be adduced in which error results from our trusting the operation of the mind more than sense” (HR 2.253: AT 7.439).

48

reflection and the stability of belief

g; a cognitive faculty is epistemologically basic just in case there is no cognitive faculty epistemologically prior to it. The modal force of the notion that a belief based on one cognitive faculty “can be corrected” or is “correctable” by another cognitive faculty needs to be clarified. To say, for example, that beliefs based on senseperception can be corrected by reason does not mean that reason in fact has the resources to show that every belief based on sense-perception is false. If that were the case, reason would have the resources to show that the belief that there exists a material world is false, whereas Descartes holds that this belief is true. The idea has to be that it follows from the nature of reason and sense-perception qua cognitive faculties that reason is not precluded from correcting beliefs based on sense-perception, whereas sense-perception is precluded from correcting beliefs based on reason. Consider the example of belief in the existence of a material world. The fact that reason does not have the resources to correct this belief is not a consequence of the nature of reason and sense-perception. The belief in the existence of a material world is correctable in the sense that reason might have located a disproof of this belief. For example, reason might have produced a proof of the existence of an omnipotent deceiver; or reason might have produced a proof, in the style of Berkeley, to show that the notion of a material substance is incoherent. Descartes’ position is that reason is epistemologically prior to senseperception, and indeed epistemologically basic, in the sense defined. The textual evidence I have produced for this claim derives from the Rules (where Divine veracity plays no role)33 and from passages in Meditation III prior to the proofs of the existence of a nondeceiving God.34 This is significant. It means that Descartes’ commitment to the epistemological position I have attributed to him is not grounded in the proof of the nondeceiving God. The claim that reason is epistemologically basic, and hence epistemologically prior to sense-perception in particular, is not itself deduced as a consequence of Divine veracity; it has the status of a purported necessary truth about the interconnections among various cognitive faculties. I want to emphasize that this body of doctrine is a component of Descartes’ epistemology even if the dissimulation hypothesis is false. Let us stipulate that the output of one’s hierarchically ordered set of cognitive faculties consists in those beliefs which have withstood or survived all possible tests for correction. We can say that for Descartes a belief is maximally reasonable if it is included within the output of one’s hierarchically

33. The following passage is perhaps suggestive of the doctrine: “For the human mind has in it something that we may call divine, wherein are scattered the first germs of useful modes of thought” (HR 1.10: AT 10.373). Descartes writes more neutrally, however, of “certain primary germs of truth implanted by nature in human minds” (HR 1.12: AT 10.376). 34. I refer to the textual evidence produced in the body of this chapter, not to the additional evidence in note 32. The latter derives from the Sixth Replies. Even these passages, however, make no reference to Divine veracity.

dissimulation in descartes’ MEDITATIONS

49

ordered set of cognitive faculties.35 This is the core of the nontheological epistemological theory. Since reason is an epistemologically basic cognitive faculty, such that beliefs based on reason are not correctable by any other faculty, beliefs based on reason are ipso facto maximally reasonable. Since sense-perception is not an epistemologically basic cognitive faculty, beliefs based on sense-perception are maximally reasonable only if they have been subjected to and withstood all possible tests for correction. Descartes does appeal to the existence of a nondeceiving God in order to validate the output of the hierarchically ordered set of cognitive faculties. In the Second Replies, Descartes writes: [I]n the case of our clearest and most accurate judgments which, if false, could not be corrected by any that are clearer, or by any other rational faculty, I clearly affirm that we cannot be deceived. For, since God is the highest being He cannot be otherwise than the highest good and truth, and hence it is contradictory that anything should proceed from Him that positively tends toward falsity. (HR 2.40–41: AT 7.143–44) The mere fact that we possess a faculty that leads to false beliefs does not render God a deceiver. God is a deceiver only if we possess a faculty that leads to false beliefs which are not correctable by other faculties we possess. This is explicit in Meditation VI, where Descartes writes that it follows from “the sole ground that God is not a deceiver” that “He has not permitted any falsity to exist in my opinion which he has not likewise given me the faculty of correcting” (HR 1.191; AT 7.80).36 In Meditation VI, Descartes does appeal to Divine veracity in order to guarantee the truth of those beliefs which withstand or survive all possible tests for correction. From the perspective of the dissimulation hypothesis, Descartes does not take the theological validation of the hierarchically ordered set of cognitive faculties seriously. The dissimulation hypothesis treats the Divine validation as a cosmetic graft onto a relatively healthy epistemological theory. The theory of hierarchically ordered cognitive faculties embodies a nontrivial, substantive epistemological position. Furthermore, it is a theory to which Descartes commits himself independently of any appeal to Divine veracity. 35. We can identify Cartesian knowledge with those beliefs which are both included in the output of one’s hierarchically ordered set of cognitive faculties, and true. This raises an important question, which I hope to address elsewhere: in the absence of any appeal to Divine veracity, is there any Cartesian argument to show that beliefs included within the output of one’s hierarchically ordered set of cognitive faculties are likely to be true? 36. Descartes writes in Meditation IV: “as He could not desire to deceive me, it is clear that He has not given me a faculty that will lead me to err if I use it aright” (HR 1.172: AT 7.54). This passage is not inconsistent with the position in Meditation VI, provided that the notion of using a faculty correctly is construed to include subjecting the faculty to correction by other faculties, as appropriate. The passage need not be interpreted to mean that the correct use of a faculty, in and of itself, leads only to true beliefs.

50

reflection and the stability of belief

Locating this theory in Descartes enables us to meet the objection that in the absence of any appeal to Divine veracity Descartes’ epistemology virtually vanishes. What, then, remains of the argument for the existence of the material world in Meditation VI? The argument is embedded within a long discussion of “the teachings of nature.”37 In Meditation III, Descartes has already contrasted the teachings of nature with the light of nature, identifying the former with “a certain spontaneous inclination which impels me to believe” (HR 1.160: AT 7.38). At the fifth paragraph of Meditation VI, Descartes begins to discuss a particular class of beliefs: beliefs involving “those matters which I hitherto held to be true, as having perceived them through the senses” (HR 1.187: AT 7.74); beliefs held when “I had formerly made use of my senses rather than my reason” (HR 1.188: AT 7.75); beliefs held at a time when “all the faith which I had rested in my senses” (HR 1.189: AT 7.76). These are the beliefs that had previously been based on sense-perception alone. Descartes writes that these beliefs were taught by nature or learned from nature (HR 1.188: AT 7.76). At paragraph seven, Descartes states that “nature seemed to cause me to lean towards many things from which reason repelled me” (HR 1.189: AT 7.77). We have already encountered an example in beliefs about the size of astronomical bodies. Of those beliefs which arise spontaneously on the basis of sense-perception, which ought to be accepted, and which rejected? On the epistemological theory I have located in Descartes, we should expect that beliefs based on sense-perception should be accepted just in case they have been subjected to and withstood all possible tests for correction. Some beliefs based on senseperception would survive this process, and others would not. Descartes thus writes: “I do not in truth think that I should rashly admit all the matters which the senses seem to teach us, but, on the other hand, I do not think that I should doubt them all universally” (HR 1.189–90: AT 7.77–78). Descartes proceeds case by case. That there exists a material world at all is proved as follows: But, since God is no deceiver, it is very manifest that He does not communicate to me these ideas immediately and by Himself, nor yet by the intervention of some creature in which their reality is not formally, but only eminently, contained. For since He has given me no faculty to recognise that this is the case, but, on the other hand, a very great inclination to believe [that they are sent to me or] that they are conveyed to me by corporeal objects, I do not see how He could be defended from the accusation of deceit if these ideas were produced by causes other than corporeal objects. Hence we must allow that corporeal things exist. (HR 1.191: AT 7.79–80) 37. The teachings of nature play a more prominent role in Descartes’ epistemology than commentators generally recognize. For two notable exceptions, see Stout 1932, esp. 191–97, and Kemp Smith 1952, 247–58, 286–93.

dissimulation in descartes’ MEDITATIONS

51

It is not simply because one has a very great inclination to believe that material objects cause sensory experiences that the Divine veracity guarantees that a material world exists. It is because one has this inclination and has no faculty to recognize that this belief is false, that is, no faculty to correct this belief. The Divine guarantee does not apply to any arbitrary belief arising from sense-perception. Such a guarantee would be much too blunt an epistemological instrument. The Divine guarantee applies only to beliefs that are resistant to correction. It should be noted that the conclusion of Descartes’ argument that some corporeal or material objects exist is quite weak. In the analogous passage in the Principles of Philosophy, Descartes writes: “But God cannot deceive us. . . . And hence we must conclude that there is an object, extended in length, breadth, and depth” (HR 1.254–55: AT 8A.41). We do not yet have an argument for any detailed conclusions about the material world; for example, that particular bodies exist, or that Descartes has a body. Descartes proceeds to accept a number of other beliefs that arise spontaneously on the basis of sense-perception. These include the beliefs: “that I have a body which is adversely affected when I feel pain” (HR 1.192: AT 7.80); “that I am not only lodged in my body as a pilot in a vessel, but that I am very closely united to it” (HR 1.192: AT 7.81); and “that many other bodies exist around mine” (HR 1.192: AT 7.81). Descartes does not claim that these beliefs should be accepted because they are clearly and distinctly perceived to be true. These beliefs are not based on the light of nature; they are teachings of nature, beliefs arising from sense-perception. In the case of each of these beliefs, if the argument is to parallel that for the existence of a material world, we must attribute to Descartes the implicit claim that the belief in question would withstand all possible tests for correction.38 By contrast, a number of other beliefs based on sense-perception “contain some error” (HR 1.193: AT 7.82). Here Descartes lists the beliefs: that a space in which there is nothing that affects one is void; that qualities exactly similar to sensory experiences of secondary qualities exist in bodies; and that distant bodies have the shape and size they appear to have (HR 1.193: AT 7.82). In each case, Descartes is content to provide the briefest indication of the considerations available to reason which correct the beliefs in question (HR 1.193–94: AT 7.82–83). A complete argument for the claim that the relevant beliefs should be rejected would have to include a fuller statement of these considerations.39 38. It is true that Descartes provides virtually no argument for these claims. For example, in the central case of belief in the existence of a material world, Descartes asserts, rather than argues, that this belief is resistant to correction. This is a defect in the Meditations whether or not one takes seriously the appeal to Divine veracity. As I have emphasized earlier, the Divine veracity at best guarantees the truth of those beliefs which are resistant to correction. On any interpretation, Descartes ought to have done more to show that the beliefs validated in Meditation VI satisfy this condition. 39. Such statements are forthcoming in other writings. For example, Descartes provides an extended argument against the possibility of a vacuum in Principles II.10–18.

52

reflection and the stability of belief

The important point is that the appeal to Divine veracity does not in itself enable Descartes to discriminate among the beliefs that arise spontaneously on the basis of sense-perception. The acceptability of each such belief depends crucially on whether or not the belief is resistant to correction. It is the claim that particular beliefs would, or would not, withstand all possible tests for correction which bears the argumentative weight. From the perspective of the dissimulation hypothesis, Descartes was not serious about the theological validation of the output of the hierarchically ordered set of cognitive faculties. Our discussion shows that if we jettison the appeal to Divine veracity in Meditation VI, the core of a Cartesian argument for the truth or falsity of various beliefs induced by nature is not lost. The heart of the argument would consist, in each instance, in an attempt to show that the belief in question would or would not withstand all possible tests for correction. It will be objected that something has been lost, namely, the Divine validation of those beliefs which are resistant to correction. From the perspective of the dissimulation hypothesis, what is “lost” is something that Descartes did not believe could be won. According to this hypothesis, Descartes was aware that, in the face of doubt about the reliability of clear and distinct perception, it would be question-begging to validate beliefs based on clear and distinct perception by appeal to Divine veracity, where the proof of the existence of a nondeceiving God itself relies on clear and distinct perception. Similarly, Descartes would have realized that, in the face of doubt about the reliability of the hierarchically ordered set of cognitive faculties, it would be question-begging to validate beliefs based on this hierarchically ordered set by appeal to Divine veracity, where the proof of the existence of a nondeceiving God itself relies on the hierarchically ordered set of cognitive faculties. It is no more possible, without circularity, to validate the hierarchically ordered set of cognitive faculties by proving the existence of a nondeceiving God than it is to validate clear and distinct perception itself by proving the existence of a nondeceiving God.40

40. Of course, it would not have been question-begging to prove the existence of a nondeceiving God via an argument (whether the arguments of Meditation III or the ontological argument of Meditation V) which relies exclusively on clear and distinct perception in order to validate sense-perception. In other words, the Meditations would not have been circular had it been different in two respects: first, had the doubt about clear and distinct perception itself not been introduced in Meditation III and, second, had Descartes appealed to Divine veracity solely for the purpose of validating cognitive faculties other than clear and distinct perception. It is the fact that Descartes wrote and published the Meditations, rather than the variant of it I have described, which leads me to think that the dissimulation hypothesis is more likely to be true than its weakened version (see note 10). I suggest in §4 that Descartes’ motive for dissimulation was to render his physics more acceptable to the church by proposing a fundamental epistemological role for God. In the variant of the Meditations I have described, knowledge of the existence of God is necessary for knowledge of the existence of the material world. Thus, if the weakened version of the dissimulation hypothesis were correct, Descartes could have made this epistemological offering to the church without dissimulating at all.

dissimulation in descartes’ MEDITATIONS

53

I believe that attributing the nontheological epistemology to Descartes has some modest explanatory power. We have seen that in Meditation I Descartes declines to rely on the hypothesis that sense-perception is intrinsically defective. This can be explained with reference to the epistemological position I have sketched. We think of Descartes as identifying maximally reasonable belief with the output of the hierarchically ordered set of cognitive faculties. Various beliefs based on sense-perception alone do not withstand or survive all possible tests for correction; however, many beliefs based on sense-perception are resistant to correction. If sense-perception were intrinsically defective, beliefs about the material world which survive all possible tests for correction could be systematically false. This is, of course, a possibility, and Descartes never suggests otherwise. Descartes would not have invoked the hypothesis that sense-perception is intrinsically defective because in his own epistemological theory sense-perception plays an ineliminable, albeit correctable, role in generating reasonable belief. In the absence of Divine validation of our faculties, we must rely on the output of the hierarchically ordered set of cognitive faculties, and hence on the contributions of the component faculties, to include sense-perception. Descartes’ refusal to challenge the reliability of sense-perception can be construed as reflecting his recognition of this point. We have also seen that Descartes characterizes the doubts of Meditations I and III quite differently. The former is powerful and mature, the latter slight and metaphysical. I have noted two disanalogies between the hypotheses as presented by Descartes, disanalogies that might be thought to justify the difference in his assessment of the relative force of the doubts. The disanalogies are superficial, however, in the sense that Descartes could have constructed the two hypotheses in a way that would not have involved the disanalogies. What explains Descartes’ preference for constructing the two hypotheses in a way that enabled him to minimize the doubt of Meditation III? The dissimulation hypothesis provides an answer. Suppose Descartes did subscribe to the nontheological epistemological position I have attributed to him. Then the point of Meditation I is that one cannot have (maximally) reasonable belief about the material world on the basis of sense-perception alone, on the basis of beliefs that arise spontaneously from sensory experience and which have not been subjected to all possible tests for correction. Thus, in the third paragraph of Meditation I, Descartes writes: “All that up to the present time I have accepted as most true and certain I have learned either from the senses or through the senses” (HR 1.145: AT 7.13). The hypothesis of the omnipotent deceiver is a particularly striking way of showing that beliefs about the material world, insofar as they are based on sensory experience alone, are correctable in the sense that they are susceptible to correction by reason; there is nothing in the nature of reason qua cognitive faculty which precludes it from correcting beliefs based on sensory experience by showing, for example, that there does exist an omnipotent deceiver. The function of the hypothesis of Meditation I is to suggest that senseperception is not epistemologically basic. This is the initial stage of the

54

reflection and the stability of belief

argument for the hierarchical ordering of the cognitive faculties. The argument is completed in Meditation III when Descartes claims, prior to producing the proofs of the existence of God, that reason or the faculty of clear and distinct perception is epistemologically basic. In Meditation VI, against the background of the claim that the cognitive faculties are hierarchically ordered, Descartes restores or reinstates some of the beliefs that had previously been based on sense-perception alone, by contending that they are resistant to correction. If this account is correct, there is a legitimate function for the hypothesis of Meditation I, but there is no comparable function for an analogous hypothesis in Meditation III. The point of the hypothesis of Meditation I is that sense-perception is susceptible to correction by reason and hence not epistemologically basic. There is no corresponding function to be served by the hypothesis about reason in Meditation III. If one thinks of sense-perception and reason as epistemologically coordinate or as coequal cognitive faculties, it will seem that equally strong doubts can be raised about both. Descartes, however, held that reason or the faculty of clear and distinct perception is epistemologically prior to sense-perception, and indeed epistemologically basic. There can be no question of this faculty’s being susceptible to correction by some yet more basic cognitive faculty. This is the underlying rationale for treating the doubt about reason as slight and metaphysical, whereas the doubt about sense-perception is treated as powerful and mature. A full, direct explanation of this rationale would defeat the dissimulation by making the nontheological epistemology explicit. Descartes prefers to construct the hypotheses of Meditations I and III in a way that renders them superficially disanalogous. These disanalogies, which provide superficial reasons for the disparate characterizations of the grounds for doubt in Meditations I and III, function as proxies for the underlying rationale I have described. Attributing the nontheological epistemology to Descartes would enable us to understand why Descartes adopted such different postures toward the hypotheses of Meditations I and III. According to the dissimulation hypothesis, Descartes nevertheless introduces the hypothesis of Meditation III as a pretext for locating a particular epistemological role for God, and it is at this juncture that dissimulation comes to dominate the developments of Meditations III and IV. In Meditation III, Descartes offers arguments for the existence of a nondeceiving God, even though he did not believe the arguments cogent. In Meditation IV, Descartes appeals to the existence of this being in support of his claim that whatever one clearly and distinctly perceives is true, even though he was aware of the circularity of his procedure.

4. Descartes’ Motives Why would Descartes have engaged in such massive dissimulation? I will sketch an answer to this question. Descartes began writing his Treatise on the Universe in 1629. The work was ready for publication in 1633. In June

dissimulation in descartes’ MEDITATIONS

55

of that year, Galileo was condemned by the Inquisition for his doctrine that the earth moves. This doctrine was also central to Descartes’ cosmology. When Descartes heard of the condemnation, he decided not to publish his treatise. Descartes feared censure by the church. His fear was not ill-founded, since his philosophy was condemned by Rome in 1663, sixteen years after his death. Descartes’ concerns were reinforced by his positive desire that his philosophy be taught in church colleges.41 Descartes went to great lengths in attempting to ensure that the church received the Meditations favorably. We know that the dedication to the theology faculty at the Sorbonne and the title page of the first edition misrepresented the work as containing a proof of the immortality of the soul.42 Descartes wrote to Mersenne in 1639: “To make it as good as possible, I plan to have only twenty or thirty copies printed, and send them to the twenty or thirty most learned theologians I can find, so as to have their criticisms and learn what should be changed, corrected or added before publication” (K 68: AT 2.617). He wrote to Mersenne in 1640 of his desire for “the approbation of the Sorbonne, which I want, and which seems very useful for my purposes. Because I must confess that the small Treatise on Metaphysics which I sent you contains all the principles of my Physics” (K 82: AT 3.233). It fell to Mersenne to solicit criticisms of the Meditations from theologians, though Mersenne approached others as well. It is these criticisms that constitute the various sets of Objections. Descartes misrepresented aspects of his scientific views (insofar as they related to creation and to the earth’s motion) in print in the Principles.43 In light of Descartes’ prudential concerns, it does seem possible that in the Meditations Descartes purposely misrepresented his views with respect to the role of God in epistemology. The effect of the misrepresentation is to suggest a specific, intimate connection between the existence of God and the possibility of human knowledge—that Divine validation of the cognitive faculties is necessary for human knowledge; the intent is to render other aspects of Descartes’ philosophy, most especially his physics, more acceptable to the church. I have noted that in the Rules, written in 1628–1629, there is no hint of skepticism about intuition or clear and distinct perception, and no epistemological role for God. Galileo was condemned in 1633. In the Discourse on Method, published in 1637 as a preface to treatises on light, meteors, and geometry, Descartes maintains that the certainty of clear and distinct perception is due to Divine perfection. I suggest that the apparent change in Descartes’ views about the epistemological status of clear and

41. For accounts of Descartes’ various prudential concerns, see Adam 1910, 165–79; Caton 1970, 14–19; Haldane 1905, 153–57; Mahoney 1979, ix–xiv; and B. Williams 1978, 18. 42. Cf. Adam 1910, 304; Caton 1973, 101–2; Curley 1978, 99; and Dorter 1973, 316–18. 43. For the impact of Descartes’ prudential concerns on the substance of the Principles, see Adam 1910, esp. 361–65, 372–74, 337–83; A. B. Gibson 1932, 26, 264–68; Haldane 1905, 289–90; and B. Williams1978, 371–72.

56

reflection and the stability of belief

distinct perception may have resulted from Descartes’ attempt, most especially after the condemnation of Galileo, to make his philosophy more acceptable to the church.44 I have not attempted to establish that the dissimulation hypothesis is correct; however, I think it would be incautious to reject dissimulation hypotheses out of hand.45

44. Curley advances a different explanation of the discrepancy between Descartes’ treatment of clear and distinct perception in the Rules, on the one hand, and in the Discourse on Method and Meditations, on the other (1978, 35–37). 45. This essay benefited from discussion or correspondence with Robert Adams, Miles Burnyeat, Julie Heath, David Hills, Jaegwon Kim, Amélie Rorty, and David Velleman.

dissimulation in descartes’ MEDITATIONS

57

2 The Priority of Reason in Descartes

1. Introduction according to descartes, it is reason which corrects sense-perception in cases of disagreement or conflict between the two faculties. In his reply to the ninth difficulty raised in the Sixth Objections, Descartes claims that it is reason which corrects the visual judgment that a stick protruding from water is bent. Reason also corrects perceptual judgments about the size of distant objects, such as the sun (cf. HR 1.161, 193, 106; O 111: AT 7.39, 82, 6.39–40, 144).1 When Descartes writes of one faculty correcting another, he uses ‘emendare’ and ‘corriger’ (AT 7.80, 89, 144, 439, 9A.64, 71, 113, 238). I therefore take it that Descartes’ doctrine about correction is normative: whereas the proper use of reason does not require submitting the beliefs it generates to tests for correction by sense-perception, the proper use of senseperception does require submitting the beliefs it generates to tests for correction by reason. (I document Descartes’ commitment to this doctrine later.2) I introduce the phrase ‘priority of reason’ stipulatively, as a label for the italicized doctrine. According to this doctrine, reason and sense-perception are hierarchically ordered. Throughout this chapter, I use the term ‘reason’ to refer to the faculty of clear and distinct perception.3 1. [In this chapter, quotations of Hume are from the Selby-Bigge and Nidditch editions of the Treatise and first Enquiry.] 2. See the third paragraph that follows and note 5. 3. My interpretive claims about this faculty rely on passages where Descartes uses the terms ‘ratio’ or ‘raison’, ‘intellectus’ or ‘entendement’, ‘mens’ or ‘esprit’, and ‘lumen naturalis’ or ‘lumière naturelle’. When it seems required, I provide textual justification for interpreting these passages as applicable to the faculty of clear and distinct perception.

58

Any interpretation of Descartes’ epistemology must explain his grounds for the normative claim that, in cases of conflict, it is reason which corrects sense-perception. What is to rule out the possibility that it is sense-perception which corrects reason? What is to rule out the possibility that reason and sense-perception are coordinate faculties, and that in cases of conflict either they trade off until they reach reflective equilibrium or they deadlock in ineliminable incoherence? The priority of reason must be grounded in some asymmetry between reason and sense-perception. In this chapter, I attempt to identify the asymmetry that ultimately serves this function. In order to demonstrate that there is an interpretive problem here, I turn to an obvious, but I think futile, strategy for explaining Descartes’ commitment to the priority of reason. The strategy is to exploit Descartes’ commitment to the “truth rule” in order to ground the priority of reason. According to the truth rule, whatever one clearly and distinctly perceives is true; clear and distinct perception is an infallible source of true belief. Descartes does not subscribe to an analogous “truth rule” for sense-perception; he does not maintain that whatever one believes on the basis of sense-perception is true. This suggests that the priority of reason is grounded in the superiority of reason as a source of true belief, in the greater truth-conduciveness of reason: whereas reason is infallible, sense-perception is fallible. According to the strategy under consideration, the priority of reason ultimately rests on the truth rule and the greater truth-conduciveness of reason. The difficulty is that Descartes’ argument for the truth rule itself relies on the priority of reason. Descartes holds that the truth rule is a consequence of the existence of a nondeceiving God. In the version of the argument in the Second Replies, Descartes claims, on the ground that “it is contradictory that anything should proceed from [God] that positively tends toward falsity,” that “in the case of our clearest and most accurate judgments which, if false, could not be corrected by any that are clearer, or by any other natural faculty, I clearly affirm that we cannot be deceived” (HR 2.40–41: AT 7.143–44). If clear and distinct perception were susceptible to correction by some higher faculty, the mere fact that clear and distinct perception led to false belief would not render God a deceiver—God would be a deceiver only if the false belief withstood all appropriate tests for correction. The inference from the existence of a nondeceiving God to the truth rule requires the independent assumption that there is no faculty for correcting judgments based upon clear and distinct perception, that the proper use of reason does not require submitting the beliefs it generates to tests for correction by senseperception, or indeed by any other faculty.4 The italicized portion of this

4. The application of the infallibility rule to clear and distinct perception at the final paragraph of Meditation IV does not make the presupposition explicit. In this respect, the argument for the truth rule in the Second Replies is a more careful statement of Descartes’ position. In the Meditations, however, the presupposition has already been introduced at paragraph nine of Meditation III: “I cannot doubt that which the natural light causes me to believe to be true, as, for example, it has shown me that I am from the fact that I doubt. . . . And I possess no other

the priority of reason in descartes

59

assumption is identical to (the first half of) the priority of reason.5 Since the priority of reason figures in the explanation of the truth rule, the priority of reason must have some basis that is independent of the truth rule. (I develop this point in more detail in §5.) This leaves us with the question of what explains Descartes’ commitment to the priority of reason. My thesis is that the priority of reason to senseperception ultimately rests on the greater irresistibility of reason. According to my interpretation, Descartes holds that whereas reason generates beliefs that are psychologically irresistible, sense-perception generates suppressible inclinations to believe; Descartes also holds that permanence is a doxastic objective; thus, in the pursuit of permanence in belief, one need not submit reason to tests for correction by sense-perception, but one ought to submit sense-perception to tests for correction by reason. Some of the elements of this interpretation are doubtless familiar. Descartes’ doctrine of the irresistibility of reason is entrenched in the literature.6 The theme of permanence is an important element in a number of recent discussions of the problem of the Cartesian circle.7 My claim is that a psychological asymmetry between reason and sense-perception, together with the adoption of permanence as a doxastic objective, explains the priority of reason. In §§2 and 4, I explore interpretations of Descartes’ grounds for asserting the priority of reason that are different from my own. I develop my positive interpretation in §3. In §5, I explain how Descartes could exploit his doctrine of the greater irresistibility of reason in order to establish the greater truth-conduciveness of reason. In §6, I consider the bearing of the problem

faculty whereby to distinguish truth from falsehood, which can teach me that what this light shows me to be true is not really true, and no other faculty that is equally trustworthy” (HR 1.160–61: AT 7.38–39). It is clear from the use of the cogito as an example that the natural light includes clear and distinct perception (cf. HR 1.7, 158: AT 10.368, 7.35). The claim at paragraph nine is therefore that reason is not susceptible to correction by any other faculty. 5. The second half of the priority of reason emerges in the course of Descartes’ application of the infallibility rule to sense-perception, and to belief in the existence of a material world in particular. Descartes writes in Meditation VI that “God is not a deceiver, and . . . consequently He has not permitted any falsity to exist in my opinion which He has not likewise given me the faculty of correcting” (HR 1.191: AT 7.80). Thus, at the preceding paragraph of Meditation VI, the argument that a material world exists depends on three claims: that Descartes has a strong inclination to believe that the material world exists, that God is no deceiver, and that “He has given me no faculty to recognise” (HR 1.191: AT 7.79) that ideas of material objects are caused by something other than material objects. The point of the latter claim is that the belief that ideas of material objects are caused by material objects “passes” tests for correction by other faculties, and by reason in particular. This is an essential step in the argument only on the assumption that the proper use of sense-perception requires submitting the beliefs it generates to tests for correction by reason. Since sense-perception is susceptible to correction by a higher faculty, the mere fact that sense-perception led to false belief would not render God a deceiver; God would be a deceiver only if the false belief withstood all appropriate tests for correction. 6. Cf., for example, Kenny 1968, 179–82; Frankfurt 1970, 154, 163–69; Tlumak 1978, 54; and B. Williams 1978, 180, 183–87, 306, and 1983, 345–46. 7. See Kenny 1968, 192–93; Frankfurt 1970, 24, 124, 179–80; Tlumak 1978, 45–49, 58–63; B. Williams 1978, 202–4, and 1983, 345, 349; and Rodis-Lewis 1986, 277–81.

60

reflection and the stability of belief

of the Cartesian circle on my account of the priority of reason. In §7, I argue that the epistemological theory that emerges in §3 has interesting affinities with the epistemology of Hume.

2. The Coherence of Reason In this section, I explore the possibility that the priority of reason is grounded in the claim that reason is internally coherent, and sense-perception internally incoherent. Descartes’ commitment to an asymmetry with respect to coherence emerges in the Objections and Replies. Descartes writes, in the Sixth Replies, of the case in which refraction makes a stick protruding from water appear bent: But I cannot grant what you here add, viz. that that error is corrected not by the understanding but by the touch. For, although it is owing to touch that we judge that the staff is straight, . . . this, nevertheless, does not suffice to correct the error. Over and above this we need to have some reason to show us why in this matter we ought to believe the tactual judgment rather than that derived from vision; and this reason . . . must be attributed not to sense but to the understanding. Hence in this instance it is the understanding solely which corrects the error of sense; and no case can ever be adduced in which error results from our trusting the operation of the mind more than sense. (HR 2.253: AT 7.439) Descartes calls attention to the fact that perceptual judgments sometimes conflict.8 He claims that there are no resources internal to sense-perception sufficient to resolve the conflict, and that the conflict can be resolved by reason. Reason would not have the resources to resolve the conflict if it were itself incoherent. Descartes must therefore hold that reason does not exhibit internal conflicts analogous to those found within sense-perception. This thesis emerges more explicitly in the Second Replies, in the reply to the fourth objection in the Second Objections. In the ninth paragraph of this reply, Descartes introduces the example of jaundice: In matters perceived by sense alone, however clearly, certainty does not exist, because we have often noted that error can occur in sensation, as . . . when one who is jaundiced sees snow as yellow; for he sees it thus with no less clearness and distinctness than we see it as white. If, then, any certitude does exist, it remains that it must be found only in the clear perceptions of the intellect. (HR 2.42: AT 7.145) 8. Descartes’ claims that error can occur in sensation, and that visual judgments can conflict, are to be understood against the background of his distinction between three grades of sensation in the Sixth Replies. Any error in sensation requires judgment, and judgment is involved only in the third grade of sensation (cf. HR 2.251–52: AT 7.436–38).

the priority of reason in descartes

61

Once again, Descartes claims that reliance on sense-perception on its own leads to ineliminable incoherence.9 He concludes that there is no certainty based on sense-perception on its own, and that certainty must either be achieved through the use of reason, or not at all. In the tenth paragraph, Descartes maintains that some perceptions of reason are indeed certain. In the eleventh paragraph, he considers an objection: No difficulty is caused by the objection that we have often found that others have been deceived in matters in which they believed they had knowledge as plain as daylight. For we have never noticed that this has occurred, nor could anyone find it to occur with these persons who have sought to draw the clearness of their vision from the intellect alone, but only with those who have made either the senses or some erroneous preconception the source from which they derived that evidence. (HR 2.42: AT 7.146) Descartes claims that deception could not occur through the use of reason on its own. In context, the point is that reason is not susceptible to the same kind of uncertainty that infects sense-perception, and hence that reason does not admit of internal incoherence.10 Descartes’ strategy in defending this position is to explain conflicts that apparently arise within reason as due to reason’s contamination either by sense-perception or by a false presupposition.11 Descartes maintains that no conflict could occur from the use of reason on its own, that is, from the use of pure reason. In the reply to the fourth objection, Descartes discusses the intellect— ‘intellectus’ at AT 7.145, 146 and ‘l’entendement’ at AT 9A.114. I have taken the discussion to apply to reason in the sense of §1, the faculty of clear and distinct perception. One of Descartes’ examples of a perception of the

9. Although Descartes is fond of the example of jaundice (cf. HR 1.44, 105; O 110: AT 12.423, 6.39, 142), the example of the protruding stick has the advantage that the perceptually generated conflict is internal to a single cognizer, who possesses both an inclination to believe that the stick is bent and an inclination to believe that the stick is straight. (An object’s feeling warmer to one hand than to the other would constitute an analogue to the case of the stick involving a single sense.) By contrast, the jaundiced person need not have a perceptually generated conflict since he need not, on the basis of perceptual experience, have any inclination to believe that snow is white. He may have been jaundiced from birth. 10. This claim emerges in the early work the Rules for the Direction of the Mind: “By intuition I understand, not the fluctuating testimony of the senses . . . but the conception which an unclouded and attentive mind gives us so readily and distinctly that we are wholly freed from doubt about that which we understand” (HR 1.7: AT 10.368). The fluctuating testimony of the senses just is the manifestation over time of internal incoherence. Reason, by contrast, does not exhibit such fluctuations. 11. The Rules contains a mitigated version of this claim: “it is a matter of experience that the most ingenious sophisms hardly ever impose on anyone who uses his unaided reason” (HR 1.32: AT 10.406). Translating ‘pura’ as “unadulterated” or “pure” would better convey what Descartes intends than “unaided” as supplied by Haldane and Ross. The point is that the sophisms fail to convince reason when it is uncontaminated.

62

reflection and the stability of belief

intellect at paragraph ten of the reply is “that I, while I think, exist” (HR 2.42: AT 7.145). Knowledge of one’s own existence is attributed to clear and distinct perception both in the Meditations and in the Rules for the Direction of the Mind (HR 1.7, 158: AT 10.368, 7.35). In the French version of the Second Replies, the discussion is explicitly directed at what the intellect “clearly and distinctly conceives” (AT 9A.114). The claim that the intellect is internally coherent is intended to entail that clear and distinct perception is internally coherent. Descartes holds that whereas (pure) reason is internally coherent, senseperception is not.12 This asymmetry with respect to coherence can explain why it is reason which corrects sense-perception in the following case: (i) sense-perception generates conflicting beliefs, one of which conflicts with reason. Since reason is coherent, only one of the beliefs generated by senseperception will conflict with reason, and thus reason will have resources to resolve the conflict. These considerations, however, do not establish, in its full generality, the thesis that in cases of conflict it is reason which corrects sense-perception. The difficulty is that Descartes recognized a second case: (ii) sense-perception generates a belief that does not conflict with any other belief based on senseperception, but which conflicts with reason. Descartes provides examples in Meditation VI: the belief in the existence of vacuums, the belief that bodies possess qualities exactly similar to sensory experiences of secondary qualities, and the belief that pain and pleasure exist in bodies (HR 1.192–94: AT 7.82–83). It is important that case (ii) does not reduce to case (i). There are extended arguments, at Principles of Philosophy II.10–18 and I.66–70, culminating in the correction of the beliefs about vacuums, and about secondary qualities and sensations, respectively. Inspection of the arguments shows that they do not draw on any inclinations generated by senseperception to disbelieve that vacuums exist, or that bodies possess qualities exactly similar to sensory experiences of secondary qualities, or that sensations exist in bodies. If they did so, the present example would reduce to an instance of case (i). We are, instead, confronted with a conclusion of (pure) reason that conflicts with a belief generated by sense-perception. In case (ii), where the only conflict is between sense-perception and reason, the internal incoherence of sense-perception does not come into play.13 12. There are passages in the Discourse and the Principles that might seem at odds with my interpretation: “there are men who deceive themselves in their reasoning and fall into paralogisms, even concerning the simplest matters of geometry” (HR 1.101: AT 6.32); “We shall also doubt . . . even of the demonstrations of mathematics. . . . One reason is that those who have fallen into error in reasoning on such matters, have held as perfectly certain and self-evident what we see to be false” (HR 1.220: AT 8A.6). These passages are consistent with those in the Second Replies and the Rules, provided we suppose that the mathematical errors occur only when reason is adulterated by the senses or by some erroneous preconception. 13. Hatfield points out that in the case of the stick protruding from water, the visual judgment is corrected not by the sense of touch, but by the understanding. He comments that if this “were all that Descartes’ claim of the greater certitude of the intellect over the senses amounted

the priority of reason in descartes

63

This point would apply to any attempt to ground the priority of reason that stresses examples of mistaken perceptual judgments involving the size of distant objects (the sun), perceptual illusions (the stick protruding from water), or defective perceptual apparatus (jaundice). These are examples of the relativity of perception due to the (external or internal) conditions of observation; they involve internal conflicts within sense-perception and hence represent instances of case (i). Thus, a number of potential accounts of why Descartes takes sense-perception to be defective are subsumed under accounts with reference to the asymmetry with respect to coherence, and succumb to the difficulty I have mentioned. The asymmetry with respect to coherence cannot provide any direct rationale for holding that it is reason which corrects sense-perception in case (ii). It might be replied that it can provide an indirect rationale: if senseperception sometimes generates conflicting beliefs, and reason never does so, then we should always prefer reason to sense-perception. The strategy here is to provide a boot-strapping rationale that attempts to extend the results for case (i) in the interest of achieving generality in the overall account of the relationship between reason and sense-perception. There is a line of resistance even to the indirect rationale. This line begins with the earlier observation that the conflicts in case (i) are due to relativity in the conditions of observation. Perceptual relativity plays no obvious role in the generation of the false beliefs in case (ii). The incoherence in case (i) can therefore be isolated and prevented from infecting case (ii). This is suggestive of some difference in the sorts of cognitive processes that operate in cases (i) and (ii). According to the proposed line of resistance, “sense-perception” might even be conceived as consisting of two faculties: the cognitive processes that account for perceptual relativity and generate the conflicting beliefs in case (i); and the cognitive processes that generate false, but not conflicting, beliefs about vacuums, secondary qualities, and sensations, in case (ii). The motivation behind the indirect rationale is to achieve generality in the account of the relationship between the faculties of reason and sense-perception. This motivation seems less compelling if the beliefs attributed to “sense-perception” in cases (i) and (ii) are conceived as resulting from the operation of two different faculties. Pending a resolution of the issues raised by this line of resistance, the indirect rationale for the claim that it is reason which corrects sense-perception in case (ii) cannot be rejected out of hand. Even if the indirect rationale proves viable, however, it would be preferable to locate a rationale for the priority of reason that applies directly to both cases (i) and (ii). In the section that follows, I show that the resources for a rationale of this sort are readily available within the Meditations.

to, it would be a rather unexceptional doctrine” (1986, 59). Hatfield does not explain why an “unexceptional” doctrine cannot ground the priority of reason. My position is that it fails to do so because it does not generalize to case (ii).

64

reflection and the stability of belief

3. The Irresistibility of Reason In this section, I argue that the priority of reason can be explained by an asymmetry with respect to the psychological properties of the cognitive faculties. Descartes holds that correction is a psychological process that results in a belief’s being undermined or subverted. In his discussion in Meditation III of beliefs about the size of the sun, Descartes writes: I find, for example, two completely diverse ideas of the sun in my mind; the one derives its origin from the senses . . . ; according to this idea the sun seems to be extremely small; but the other is derived from astronomical reasonings, i.e. is elicited from certain notions that are innate in me, or else it is formed by me in some other manner; in accordance with it the sun appears to be several times greater than the earth. These two ideas cannot, indeed, both resemble the same sun, and reason makes me believe that the one which seems to have originated directly from the sun itself, is the one which is most dissimilar to it. (HR 1.161: AT 7.39) Descartes’ claim is not that reason shows that the belief based on senseperception is false; rather, reason makes him believe (‘ratio persuadet’ at AT 7.39; ‘la raison me fait croire’ at AT 9A.31) that it is false. Similarly, at paragraph seven of Meditation VI, Descartes writes that nature leads him to many beliefs from which reason “dissuades” (‘dissuadere’ at AT 7.77) him or “turns him away” (‘détourner’ at AT 9A.61). In these examples, a result of the process of correction is that the belief based on sense-perception is subverted by reason. Psychological characterizations of the relationship between the deliverances of reason and sense-perception are also given some prominence at paragraph fifteen of Meditation VI (HR 1.193–94: AT 7.83, 9A.66). The psychological features of correction in these examples are consequences of a body of psychological theory about the cognitive faculties. One element of the theory is that reason generates beliefs that are irresistible. According to Descartes, clear and distinct perception is psychologically compelling or irresistible in the sense that the belief that p is true is irresistible at any time p is clearly and distinctly perceived. He writes, for example: “I am of such a nature that as long as I understand anything very clearly and distinctly, I am naturally impelled to believe it to be true” (HR 1.183: AT 7.69; cf. HR 1.160, 236, 2.41: AT 7.38, 8A.21, 7.144).14 14. There are other passages that stress the irresistibility of clear and distinct perception without explicitly committing Descartes to the claim that the belief that p is true is irresistible at any time p is clearly and distinctly perceived. Descartes writes, for example: “our mind is of such a nature that it cannot help assenting to what it clearly conceives” (K 73: AT 3.64; cf. K 149; C 6: AT 4.115–16, 5.148). Here, the explicit claim is that the belief that p, rather than the belief that p is true, is irresistible. I am not suggesting that Descartes distinguishes the belief that

the priority of reason in descartes

65

A second element of the theory is the claim that sense-perception on its own produces inclinations to believe. Descartes identifies beliefs based solely on sense-perception with beliefs that are taught or learned by “nature,” as distinct from the light of nature.15 At paragraph nine of Meditation III, he holds that sense-perception or nature produces an inclination (‘impetus’ at AT 7.38, ‘inclination’ at AT 9A.30) to believe.16 For example, nature produces an inclination (‘propensionam’ at AT 7.80, ‘inclination’ at AT 9A.63) to believe that sensory ideas are caused by corporeal objects. In cases in which reason subverts sense-perception, conflicts between reason and senseperception must result in suppressed inclinations to believe. It follows that at least some of the inclinations generated by sense-perception are suppressible by reason. In sum, Descartes holds that there is a psychological asymmetry between reason and sense-perception: reason generates beliefs that are psychologically irresistible, whereas sense-perception generates suppressible inclinations to believe. The psychological theory not only explains the ability of reason to subvert sense-perception, but indeed has the consequence that reason will subvert sense-perception in any case of conflict between them. Suppose sense-perception and reason generate conflicting beliefs. Sense-perception on its own produces a suppressible inclination to believe that p. Reason generates a belief that ~p is true. Any use of reason has clear and distinct perception as component steps. Descartes holds that if the premises of a continuous demonstration are themselves irresistible, the conclusion is irresistible, provided one attends to the entire demonstration (cf. HR 1.184, 224: AT 7.69–70, 8A.9). The belief that ~p is true will therefore be irresistible. I assume that if the belief that ~p is true is irresistible, the belief that p is false is also irresistible. In the presence of the irresistible belief that p is false, the inclination to believe p based on sense-perception is suppressed, and the belief that p is undermined or subverted. In any case of conflict between reason and sense-perception, reason, in virtue of its irresistibility, subverts sense-perception by suppressing the inclination to believe generated by sense-perception. It should be noted that the psychological features of correction are not “merely psychological” in the sense of involving no epistemic or semantic

p and the belief that p is true. It is important, however, for the argument in the second paragraph that follows that Descartes holds that the belief that p is true is irresistible at any time p is clearly and distinctly perceived. 15. See Meditation VI, paragraphs five through eight (HR 1.187–89: AT 7.74–78). There is some tendency, as at paragraph eight of Meditation VI, to restrict the “teachings of nature” to those beliefs generated by sense-perception that survive tests for correction by reason—these are the beliefs within class (iii) discussed later in this section. Cf. Beck 1965, 257–58. 16. In the Latin edition, this is specifically a “spontaneous” inclination. I think that when Descartes describes the judgments contained in the third grade of sensation (see note 8) as ones “we have from our earliest years been accustomed to pass about things external to us” (HR 2.251: AT 7.437), his point is that these judgments are the result of spontaneous inclinations to believe.

66

reflection and the stability of belief

notions. As we have seen, where a belief that p based on sense-perception on its own is corrected by reason, reason makes one believe that p is false. Thus, the notions of truth and falsity play a role in a complete description of the content of the doxastic states that result from correction, construed psychologically. I observed at the close of §2 that the asymmetry with respect to coherence cannot provide a direct rationale for the claim that reason corrects sense-perception in case (ii), where there is no conflict internal to senseperception. The psychological asymmetry is a more promising candidate for grounding, in its full generality, the thesis that reason corrects sense-perception. This is because the psychological asymmetry applies directly to both cases (i) and (ii). Reason is always irresistible, and sense-perception always resistible; hence, the mechanism of the subversion does not depend on the presence of a conflict internal to sense-perception. But this cannot be the whole story. The priority of reason is a normative doctrine. The psychological asymmetry explains how reason in fact subverts sense-perception in cases where a conflict has been uncovered. This falls short of explaining why the proper use of sense-perception requires tests for correction by reason. An explanation is not far to seek. Beliefs based on sense-perception are susceptible to correction by reason, and correction is a psychological process of subversion; thus, beliefs based on sense-perception that are not submitted to tests for correction by reason are liable to subversion. This suggests a strategy for exploiting the psychological asymmetry in support of the priority of reason. The claim that beliefs based on sense-perception ought to be submitted to tests for correction by reason can perhaps be established with reference to permanence in belief as a doxastic objective. There is substantial evidence that Descartes did adopt this objective. Descartes’ emphasis on doxastic permanence emerges in the first paragraph of Meditation I, where he writes of establishing a “firm and permanent structure” in the sciences (HR 1.144: AT 7.17). The same theme appears in a number of passages where Descartes writes of beliefs that are “firm and immutable” (HR 2.41, 42: AT 7.145, 146), “immutable” (HR 2.245: AT 7.428), and “firm” (HR 1.314: AT 10.513); there are also contrasts with “mutable” (AT 7.69: HR 1.184) and “fluctuating” beliefs (HR 1.7: AT 10.368). I suggest that for Descartes both individual beliefs and structures of beliefs are “firm” to the extent that they are resistant to change, and that they are “permanent” to the extent that they do not change over time. Susceptibility to subversion diminishes the extent to which a belief is firm.17 Individual beliefs and structures of beliefs are “permanent” to the extent that they do not change over time. In the case of both structures of beliefs 17. I am not saying that resistance to subversion, or irresistibility, is sufficient for firmness. Descartes arguably holds that first-person beliefs about current, conscious mental states—for example, my belief at t “visually, it now appears to me as if there is wax in front of me”—are irresistible (cf. Meditation II, ¶15), but they are not permanent.

the priority of reason in descartes

67

and individual beliefs, firmness and permanence are distinct properties. Firmness, however, is conducive to permanence—the firmer a structure of beliefs, the more likely it is to be permanent. It is also a doxastic objective for Descartes that a belief system be comprehensive and indeed “complete” in the sense of including “all things that man can know” (HR 1.203–4: AT 9B.2; cf. HR 1.9: AT 10.371–72). We may suppose that Descartes would view completeness as a special case of permanence, rather than as an independent objective. Incomplete belief systems lack permanence in the sense that they are liable to change through augmentation. I can now return to the problem of providing an explanation for the priority of reason. The proper use of sense-perception presumably requires, at the ideal limit, submitting beliefs based on sense-perception to all possible tests for correction by reason.18 Submitting beliefs based on sense-perception to all possible tests for correction by reason yields beliefs that are maximally firm and hence maximally permanent, in comparison to other beliefs that sense-perception might generate. This result can be established on a case-by-case basis. In case (i), sense-perception generates conflicting beliefs, one of which would fail some test for correction by reason. Suppose one has not submitted sense-perception to a test of reason that would correct one of the conflicting beliefs. One would be left with the inclinations to hold conflicting beliefs generated by sense-perception. On the assumption that these inclinations are of equal strength, one’s doxastic system would be inherently unstable. Stability can be achieved only if one of the conflicting inclinations is suppressed as the result of failing some test of reason. It is important to take note of the role of the asymmetry with respect to coherence in this case. The significance of incoherence is that it renders sense-perception inherently unstable. (Thus, the testimony of the senses is characterized as “fluctuating”— HR 1.7: AT 10.368—in the Rules.) The significance of coherence is that it enables reason (given its irresistibility) to stabilize the doxastic system. Were reason itself incoherent, so that it conflicted with both beliefs generated by sense-perception, stability could not be achieved. In case (ii), sense-perception generates a belief that does not conflict with any other belief based on sense-perception, but which would fail some test for correction by reason. Examples are provided in §2. These beliefs constitute teachings of nature that conflict with reason. Suppose one has not submitted the belief based on sense-perception to a test of reason that would correct it. Such a belief is liable to subversion by reason and is therefore less firm than the corrected belief. This is a direct consequence of the fact that, 18. The notion of “all possible tests” for correction is ambiguous. One way to flesh out the notion would be in terms of the “availability” of various cognitive procedures to the cognizer. This requires relativizing the notion of the proper use of sense-perception to specific cognizers at specific times. The notion of “availability” has received attention in the literature on reliabilist theories of knowledge. See A. Goldman 1979, esp. 18–20, and Schmitt 1984, esp. 9–11.

68

reflection and the stability of belief

in a case of conflict, reason, in virtue of its irresistibility, subverts senseperception by suppressing the inclination to believe generated by senseperception. Once again, the rationale for holding the corrected belief is provided directly by the psychological asymmetry relative to the objective of doxastic permanence; the uncorrected belief is liable to subversion. The significance of coherence is that it enables reason (given its irresistibility) to enhance the firmness of the doxastic system. Were reason itself incoherent, it would not be able to perform this function. It might be objected that reason is liable to subversion by sense-perception, on the following grounds.19 Clear and distinct perception occurs only under favorable conditions, when the mind is “unclouded and attentive” (HR 1.7: AT 10.368). Suppose one is clearly and distinctly perceiving that p, and that sensory inputs (for example, a loud noise) then cause inattention such that one is no longer clearly and distinctly perceiving that p. In that event, it seems that sense-perception has subverted reason, and hence that beliefs generated by reason have no special claim to stability. This objection can be met by invoking a distinction between two senses in which one faculty f might subvert another faculty g: (A) beliefs generated by f undermine inclinations to believe generated by g with which they conflict, even at a time when g operates under favorable conditions; (B) the exercise of f disrupts g’s operating under favorable conditions. It is (A) which has been the focus of my discussion to this point. The objection points out that reason is susceptible to subversion by sense-perception in sense (B). Descartes would, I believe, maintain that such subversion is in principle eliminable, by sufficient practice in a proper method. The construction of a proper method is a part of Descartes’ project in the Rules and the Discourse on the Method of Rightly Conducting the Understanding. Descartes therefore emphasizes the importance of intellectual attention and concentration (see, for example, Rules IX, X, and XII).20 What is significant about subversion in sense (A) is that it can occur even when the relevant beliefs and inclinations to believe are generated by the operation of faculty g under favorable conditions. Thus, sense-perception’s susceptibility to subversion by reason in sense (A) is not eliminable even in principle, unless it is possible in principle not to exercise either sense-perception or reason at all. This, however, would violate the objective—mentioned earlier in this section— that a belief system be comprehensive. It follows that, in cases (i) and (ii), holding a belief generated by sense-perception, but that conflicts with reason, is a doxastic strategy that inevitably leads to a sacrifice in terms of either permanence or comprehensiveness. I have considered cases in which a belief generated by sense-perception conflicts with reason. There is an additional case in which the proper use of sense-perception requires submitting the beliefs it generates to all possible

19. I owe this objection to Stephen Darwall. 20. Cf. Beck 1952, 56–59; Frankfurt 1970, 150–51; and B. Williams 1983, 346–47.

the priority of reason in descartes

69

tests for correction by reason. In case (iii), sense-perception generates a belief that does not conflict with any other belief based on sense-perception and which would survive all possible tests for correction by reason. I will say that a belief that has survived all possible tests for correction by reason has been sustained by reason. Beliefs are partially sustained by reason to the extent that they pass individual tests for correction. Examples of beliefs sustainable by reason include: “the ideas of sensible things . . . are conveyed to me by corporeal objects” (HR 1.191: AT 7.79); “I have a body which is adversely affected when I feel pain”; “I am not only lodged in my body as a pilot in a vessel, but . . . I am very closely united to it”; and “many other bodies exist around mine” (HR 1.192: AT 7.80–81). These beliefs constitute teachings of nature that do not conflict with reason. Suppose one holds a belief based on sense-perception that is sustainable by reason, but which has not been submitted to all possible tests for correction. Such a belief is presumably less irresistible than the same belief would be after it had survived all possible tests for correction and thereby been sustained by reason. This is a natural extension of the explicit psychological theory. Descartes maintains that, in cases of conflict, reason, in virtue of its irresistibility, subverts sense-perception by suppressing the inclinations to believe generated by sense-perception. Here I assume that, in cases where reason partially sustains sense-perception, reason reinforces sense-perception by strengthening the inclinations to believe generated by sense-perception. Care is needed in envisioning the mechanism of reinforcement. Let p be a belief based on sense-perception that survives all possible tests for correction. The mechanism of reinforcement does not require that p is itself clearly and distinctly perceived. In that event, the belief that p would be an irresistible deliverance of pure reason. Descartes does not take the position that instances of (iii)—for example, the proposition that there exists a material world—are themselves clearly and distinctly perceived.21 Let T be a variable ranging over possible tests for correction by reason. I presume that Descartes would hold that instances of the schema “p survives T” are clearly and distinctly perceived; it is this recognition that strengthens the inclinations to believe generated by sense-perception.22 Thus, beliefs that have been sustained by reason are firmer than their unsustained counterparts. The results in cases (i–iii) yield the normative claim that the proper use of sense-perception requires submitting the beliefs it generates to all possible 21. Descartes’ statement at Principles II.1 that “we seem [videmur] to see clearly” (AT 8A.41) that our idea of matter is caused by material objects falls short of saying that we do clearly and distinctly perceive the existence of material objects. 22. Presumably, for the reinforcement to reach the maximum, it has to be the case not only that (a) for every T, reason recognizes that p survives T, but also that (b) reason recognizes (clearly and distinctly perceives) that p survives every T. Even this does not involve clearly and distinctly perceiving p itself. Of course, since T ranges over every possible test for correction, the recognition involved in (b), that p survives every T, may require a kind of knowledge that is not humanly possible. For a relevant discussion, see Frankfurt 1970, 141–43.

70

reflection and the stability of belief

tests for correction by reason. In each case, submitting a belief based on sense-perception to all possible tests for correction by reason maximizes the firmness, and hence the permanence, of beliefs based on sense-perception. Since beliefs based on reason are irresistible, the objective of permanence does not require that they be submitted to tests for correction by sense-perception. What has been constructed is a mandate or sanction for holding beliefs that result from reason, even in the absence of tests for correction by sense-perception, and for holding beliefs that result from sense-perception, when they have been submitted to all possible tests for correction by reason.23 I conclude that it is the psychological asymmetry and the associated psychological doctrines, together with the objective of doxastic permanence, which ground the normative doctrine of the priority of reason to senseperception.

4. Reason and the Recognition of Necessary Truth My claim is that a psychological asymmetry between reason and sense-perception is fundamental to the explanation of the priority of reason. I have canvassed a number of alternative explanations of the asymmetry. In §1, I rejected attempts to derive the priority of reason from the truth rule. At the close of §2, I argued that the priority of reason, in its full generality, is not grounded in Descartes’ claim that only reason is internally coherent. In this section, I explore an additional avenue for attempting to explain the priority of reason. In his “Descartes on the Consistency of Reason,” Harry Frankfurt points out that “since there is no appeal from reason to any superior natural source of knowledge, we can possess no natural basis for correcting a belief in what we have clearly and distinctly perceived.”24 I am uncertain what explanation of the priority of reason to attribute to Frankfurt.25 As far as I can see, if he 23. This mandate does not address the question of whether the resulting beliefs are true. In particular, I am not attributing to Descartes any version of a coherence theory of truth, for example, a theory on which truth consists in membership in a maximally permanent belief system. I discuss Descartes’ arguments to show that beliefs based on the proper use of the cognitive faculties are true in §§5 and 6. 24. Frankfurt 1978, 28; cf. 35. 25. There is one explanation that can be ruled out. Frankfurt cannot appeal to the asymmetry with respect to coherence to ground the priority of clear and distinct perception. Frankfurt does hold that sense-perception is internally inconsistent (1978, 33–34; 1970, 44, 170). He takes the point of the appeal to Divine veracity to be an attempt to establish that clear and distinct perception is internally consistent, and hence to establish the coherence asymmetry (1978, 34–36; 1970, 49, 176). This question is held to be pressing precisely because “we have no other faculty superior to reason, which we might invoke in an effort to resolve a conflict generated by reason itself” (1978, 35). Frankfurt’s interpretation commits him to the view that the priority of reason is established in advance of the argument for the consistency of clear and distinct perception. Frankfurt cannot hold that the coherence asymmetry is the ground of the priority of reason.

the priority of reason in descartes

71

offers an interpretation of why reason is superior to sense-perception, he does not explicitly identify it as such.26 My interest is in an explanation of the priority of reason that can be extracted from his remarks, whether or not he would endorse it, and which is distinct from the explanations I have considered to this point. Frankfurt attributes to Descartes the doctrine that clear and distinct perceptions “constitute the best testimony of our highest faculty,” that they convey “the most impeccable credentials reason can provide,” that they provide “the best reasons” for believing a proposition.27 Frankfurt’s account of the testimony, credentials, or reasons provided by clear and distinct perception is as follows: What is it, exactly, for a person to perceive something clearly and distinctly? It consists in his recognizing that the evidence he has for some proposition, or his basis in experience for accepting the proposition, is logically definitive and complete. He perceives clearly and distinctly that p when he sees that his evidence or basis for accepting p is conclusive, in the sense that it is consistent and that no body of evidence which would warrant rejecting or doubting p is logically compatible with the evidence or basis he already has.28 What is involved in a person’s recognizing that his evidence for a proposition is “logically definitive” or “conclusive”? Frankfurt writes in the paragraph following: “A person who attentively grasps the rigorous connection of the premises and the conclusion of a valid argument, for example, perceives clearly and distinctly that the conclusion must be true if the premises are true.”29 He also states that clear and distinct perception extends to “necessary truths like those of logic and mathematics.”30 It appears that Frankfurt takes clear and distinct perception to include recognition of necessary truths generally, and not merely logical truths. Frankfurt’s discussion is suggestive of the thesis that the priority of reason derives from the fact that reason is the faculty deployed in the recognition of necessary truths.31 26. He does write that “Descartes’ question [leading to the proof of a benevolent creator] is this: what basis is there for accepting what we clearly and distinctly perceive besides the irresistible conviction which having a clear and distinct perception arouses?” (1978, 32). This might be taken to imply that, prior to the proof of a nondeceiving God, the psychological asymmetry provides a basis, and indeed the sole basis, for the priority of reason. If so, we are in agreement. 27. Frankfurt 1978, 28, 36. 28. Frankfurt 1978, 28. 29. Frankfurt 1978, 28 30. Frankfurt 1978, 29; cf. Frankfurt 1970, 102–3. 31. I am aware that Frankfurt wants to extend clear and distinct perception beyond necessary truths, and that one of my quotations is embedded in a context where he makes this point: “It is not only necessary truths like those of logic and mathematics, then, which can be perceived clearly and distinctly. Logically contingent propositions may also be objects of clear and distinct perception” (1978, 29; cf. 1970, 123–24). Frankfurt’s example is the proposition “I am

72

reflection and the stability of belief

In order to evaluate this suggestion, let us take an instance of case (ii)— where there is no conflict internal to sense-perception—as an example. Let ‘v’ be the proposition that vacuums exist. Sense-perception generates an inclination to believe that v is true; reason generates the belief that v is necessarily false. If reason is correct that v is necessarily false, sense-perception is in error. But it is equally the case that if sense-perception is correct that v is true, reason is in error. The fact that reason, unlike sense-perception, generates beliefs in what it takes to be necessary truths does not ground a presumption that reason is less fallible than sense-perception. In other words, the fact that reason is directed at the recognition of necessary truths provides no grounds for thinking that reason is better at successfully recognizing necessary truths than sense-perception at successfully recognizing truths. The recognition of necessary truth is not the necessary recognition of truth.32 There is an additional difficulty for attempts to locate in the connection between reason and necessary truth an account of the priority of reason that is a genuine alternative to the explanation I have offered in §3. Such accounts of the priority of reason are at risk of collapsing into an explanation grounded in the psychological asymmetry. The pressure in this direction issues from an idiosyncratic feature of Descartes’ metaphysics, his doctrine that the “eternal truths” are dependent on God’s will. Frankfurt’s own interpretation of this doctrine exposes the pressures in question. Frankfurt holds that for Descartes “the eternal truths are inherently as contingent as any other propositions.”33 He proceeds to address a puzzling question: how are we to understand the claim that God could have made logical impossibilities, contradictions, true?34 Frankfurt offers a broad sketch of what he takes to be Descartes’ reply. The basic idea is this. We do regard

in pain” (1978, 28; cf. 1970, 134). There is an obvious strategy for subsuming first-person beliefs about current, conscious mental states under a general account of clear and distinct perception as the recognition of necessary truths. Let us say that the belief that p is incorrigible just in case it is a necessary truth that if S believes p, p is true. Although “I am in pain” is contingent, a necessary truth is lurking here. Assuming that the belief that I am in pain is incorrigible, it is a necessary truth that if I believe that I am in pain, I am in pain. Thus, Frankfurt could amend his account of clear and distinct perception (in its application to contingent, first-person beliefs about current, conscious mental states) to hold that what is clearly and distinctly perceived is not p, but rather that p is incorrigible. This would preserve the idea that clear and distinct perception is perception of necessary truth. Frankfurt adopts a similar strategy in the context of his earlier discussion of one’s certainty in the special case of one’s own existence: “[Descartes’s] conclusion can appropriately be formulated as an ascription of logically necessary truth to a certain proposition, but that proposition is not sum. It is the proposition that sum is true whenever he utters or conceives it” (1970, 101; cf. 102–3). For an interpretation that links clear and distinct perception and incorrigibility, see B. Williams 1978, 49–50, 73, 76–81, 86. 32. I am indebted to Frederick Schmitt for (whatever is good in) this formulation of the present objection. 33. Frankfurt 1977, 42. 34. Frankfurt 1977, 43.

the priority of reason in descartes

73

some propositions as necessary, and we are unable to conceive the truth of their negations. These ascriptions of necessity and impossibility derive from the occurrences of specific subjective experiences: [W]hat we identify as necessary or as impossible, depends in the end upon the occurrence of certain experiences—our experiences of an inability to refuse assent. When we now attentively consider the proposition that one and two make three, having first discriminated and analysed its terms, we cannot help assenting to it; and that is ultimately why we regard the proposition as necessary.35 That we have the experience of the inability to refuse assent when we consider some propositions is a contingent fact about our minds. Frankfurt identifies the inability to refuse assent with a compulsion to assent.36 God might have given us minds so constituted that they failed to experience a compulsion to assent when we attentively considered propositions that we now regard as necessary. We cannot conceive of the truth of the negation of a proposition we now regard as necessary, but perhaps we can conceive of our minds having been constituted in such a way that we would have regarded different propositions as necessary. How are we to understand Frankfurt’s claim that our regarding a proposition as necessary “depends in the end” on our experience of a compulsion to assent? Frankfurt writes: “The necessities human reason discovers by analysis and demonstration are just necessities of its own contingent nature.”37 The burden of Frankfurt’s discussion, I believe, is that clearly and distinctly perceiving a proposition to be a necessary truth reduces to coming to have, as the result of an appropriate intellectual process (of attention, concentration, discrimination, analysis, et cetera), a compulsion to assent, an irresistible belief that the proposition is true. What, then, are we to make of the suggestion that the priority of clear and distinct perception consists in its providing the best reasons for believing a proposition? It would seem that ultimately, or in the end, clear and distinct perception is “logically definitive” or “conclusive” only in the sense that the relevant intellectual process generates psychologically irresistible beliefs. The view that the priority of reason derives from its role in the recognition of necessary truths, when combined with Descartes’ doctrine of the eternal truths, provides a natural route to the result that the priority of reason derives from its irresistibility.38 35. Frankfurt 1977, 46. 36. Frankfurt 1977, 46, 48. 37. Frankfurt 1977, 45. 38. In generating this result, I have drawn on Frankfurt’s own interpretation of the doctrine of the eternal truths. My argument, however, is not strictly ad hominem—I am inclined to agree

74

reflection and the stability of belief

5. The Infallibility of Reason I have argued for the following interpretive claims about Descartes’ epistemology: (A) reason, unlike sense-perception, is psychologically irresistible; (B) correction is a psychological process, in the sense that whenever reason corrects sense-perception, the corrected belief is subverted; (C) permanence is a doxastic objective; (D) thus, whereas one need not submit reason to tests for correction by sense-perception, one ought to submit sense-perception to tests for correction by reason. In sum, the priority of reason is ultimately grounded in an asymmetry with respect to irresistibility, together with Descartes’ adoption of permanence as a doxastic objective. The notion of “correction” invoked at (D) is inherited from (B). What has been explained is why Descartes subscribes to a doctrine of what I will call the psychological priority of reason: whereas the proper use of reason does not require submitting the beliefs it generates to tests for subversion by sense-perception, the proper use of sense-perception does require submitting the beliefs it generates to tests for subversion by reason. (The “psychological priority of reason,” unlike the “psychological asymmetry,” is a normative doctrine.) Dissatisfaction with my interpretation will derive, I believe, from the conviction that there is a more attractive, alternative, interpretation of Descartes: (a) reason, unlike sense-perception, is an infallible source of truth; (b) correction is an epistemic process, in the sense that whenever reason corrects sense-perception, the corrected belief is false; (c) truth is a doxastic objective; (d) thus, whereas one need not submit reason to tests for correction by sense-perception, one ought to submit sense-perception to tests for correction by reason. In sum, the priority of reason is grounded in an asymmetry with respect to infallibility, together with Descartes’ adoption of truth as a doxastic objective. The notion of “correction” invoked at (d) is inherited from (b). What is explained here is why Descartes subscribes to a doctrine of what I will call the epistemic priority of reason: whereas the proper use of reason does not require submitting the beliefs it generates to tests for truth by sense-perception, the proper use of sense-perception does require submitting the beliefs it generates to tests for truth by reason. There is, however, no incompatibility between the claims in any of the lettered pairs. The claim that (A) there is an asymmetry with respect to irresistibility is compatible with the claim that (a) there is an asymmetry with respect to infallibility. Only the stronger claim that (A′) the asymmetry with

with the points of interpretation I utilize. This is not to say that they are uncontroversial. For criticism of Frankfurt’s position, see Curley 1984, 569–97, esp. 571–74, 576–83, and La Croix 1984, 455–75. I do not have space to defend the relevant aspects of Frankfurt’s interpretation. Frankfurt does not explicitly bring his own views on Descartes’ doctrine of the eternal truths to bear on his discussion of Descartes’ doctrine of the priority of reason. The discussion of the priority of reason at Frankfurt 1978 contains no reference to Descartes’ doctrine of the eternal truths, nor to the extended discussion of that doctrine at Frankfurt 1977. I doubt that it was with the present result in view that Frankfurt wrote the passage cited at note 26.

the priority of reason in descartes

75

respect to irresistibility is the only asymmetry between reason and senseperception would be incompatible with (a). The claim that (B) whenever correction takes place, the corrected belief is subverted, is compatible with the claim that (b) whenever correction takes place, the corrected belief is false.39 Only the stronger claim that (B′) correction is merely or nothing but a psychological process would be incompatible with (b). The claim that (C) permanence is a doxastic objective is compatible with the claim that (c) truth is a doxastic objective. Only the stronger claim that (C′) permanence is the only doxastic objective adopted by Descartes would be incompatible with (c). Finally, the claim that (D) the psychological priority of reason is explained by (A–C) is compatible with the claim that (d) the epistemic priority of reason is explained by (a–c). Let us look in more detail at the proposed explanation of the epistemic priority of reason. The step at (d) is the conclusion of the argument. I take (c) to be uncontroversial. The claim at (b) is itself a consequence of (a). Thus, (a) is the crucial claim in the proposed explanation. How is an asymmetry with respect to infallibility to be established? An obvious resource is Descartes’ claim that God is not a deceiver. At paragraph three of Meditation IV, Descartes provides a substantive account of what it is for an all-perfect being to be a nondeceiver: “as [God] could not desire to deceive me, it is clear that He has not given me a faculty that will lead me to err if I use it aright” (HR 1.172: AT 7.54). Let us call this the infallibility rule; it states that any faculty is infallible in the sense that it never leads to false belief, on the condition that it is used properly. Unfortunately, the infallibility rule applies to every faculty; it follows from the infallibility rule both that reason, if used properly, never leads to false belief, and that sense-perception, if used properly, never leads to false belief.40 An asymmetry with respect to infallibility has not yet been located.41

39. The notion, invoked in §2, of reason’s “resolving” conflicts that arise within senseperception also admits of psychological and/or epistemic interpretations. 40. Garber states that reason is “the faculty which, by the argument of Meditation IV, is always trustworthy if used properly” (1986, 107). It is true that, in Meditation IV, the infallibility rule is applied to reason, and not to any other faculty. But since the infallibility rule is perfectly general, it could equally have been applied to yield the conclusion that sense-perception is always trustworthy if used properly. 41. It might be suggested that the priority of reason is grounded in an asymmetry implicit in the character of the argument to establish that reason and sense-perception are both infallible, when properly used. The argument in question is an argument of reason, in the sense that all its premises are supplied by clear and distinct perception. (The argument requires the premise “I have an idea of an all-perfect God.” This is presumably established by introspection. I believe that for Descartes this would count as clear and distinct perception. The important point is that reason can validate sense-perception without the aid of sense-perception.) This observation suggests an asymmetry: whereas reason (on its own) can supply premises that entail the infallibility of sense-perception, when properly used, sense-perception (on its own) cannot supply premises that entail the infallibility of reason, when properly used. This asymmetry with respect to the method of validating the cognitive faculties cannot generate the claim that reason is prior to sense-perception in the required sense. The fact that reason can validate sense-perception,

76

reflection and the stability of belief

In order to do so, it is necessary to distinguish between the proper use of a faculty and a use of a faculty that falls short of proper use in a respect to be specified. The priority of reason is, for Descartes, embedded within a larger body of doctrine about the “proper use” of individual faculties. Consider reason in particular. The proper use of reason does not require tests for correction by sense-perception. The proper use of reason does require assenting only to those propositions that are clearly and distinctly perceived (HR 2.41: AT 7.144—cf. Meditation IV). The earlier methodological works, the Rules, and Part II of the Discourse, contain assorted precepts governing the proper use of reason—precepts about order, attention, continuous and uninterrupted movements of thought, enumerations, aids to the understanding, and so forth. (These precepts fix what it is for reason to operate “under favorable conditions,” as discussed in §3.) The priority of reason fixes one aspect of what is required by the proper use of reason. Similarly, the priority of reason fixes one aspect of what is required by the proper use of senseperception: submitting sense-perception to tests for correction by reason. Let us say that (the use of) a faculty is uncorrected just in case it has not been submitted for any tests for correction by other faculties, but has been properly used insofar as that is compatible with its not having been submitted for any tests for correction. The distinction between the proper use of a faculty and the uncorrected use of a faculty suggests the possibility of establishing an asymmetry with respect to infallibility between uncorrected reason and uncorrected senseperception. Suppose Descartes can establish that uncorrected reason is infallible. He can then establish an asymmetry with respect to infallibility as follows: uncorrected reason is infallible, that is, reason is infallible even in the absence of tests for correction; uncorrected sense-perception sometimes conflicts with uncorrected reason;42 therefore, uncorrected sense-perception is fallible. Establishing the infallibility of uncorrected reason enables Descartes to derive (a), construed as formulating an asymmetry with respect to infallibility between uncorrected faculties: uncorrected reason, unlike uncorrected sense-perception, is infallible. How is the infallibility of uncorrected reason itself to be established? The claim that uncorrected reason is infallible is equivalent to the truth rule— whatever one clearly and distinctly perceives is true. The truth rule does not state that beliefs based on clear and distinct perception are true provided they have withstood tests for correction by other faculties. The content of the truth rule is that beliefs based on uncorrected clear and distinct perception are true. I observed in §1 that Descartes’ derivation of the truth rule

but not vice versa, does not show that it is reason which corrects sense-perception in cases of conflict. 42. Uncorrected sense-perception sometimes conflicts with the proper use of reason. I show two paragraphs later that uncorrected reason coincides with the proper use of reason. It follows that uncorrected sense-perception sometimes conflicts with uncorrected reason.

the priority of reason in descartes

77

from the claim that God is no deceiver relies on (the first half of) the priority of reason. The contribution of the priority of reason to the derivation of the truth rule can be clarified against the background of the distinction between the proper use of a faculty and the uncorrected use of a faculty. Descartes takes God’s being a nondeceiver to entail the infallibility rule. Since the infallibility rule applies to the proper use of a faculty, it is the infallibility of the proper use of reason that follows directly from the infallibility rule. Descartes can in turn derive the infallibility of uncorrected reason if uncorrected reason coincides with the proper use of reason. The use of reason is uncorrected just in case it has not been submitted for any tests for correction by other faculties, but has been properly used insofar as that is compatible with its not having been submitted for any tests for correction. The assumption that the proper use of reason does not require submitting the beliefs it generates to tests for correction by sense-perception, or by any other faculty, has the consequence that the uncorrected use of reason coincides with the proper use of reason. It is this consequence that allows Descartes to derive the infallibility of uncorrected reason from the infallibility rule. The italicized portion of the assumption that leads to this consequence is (the first half of) the priority of reason. In §1, I concluded that the priority of reason must have some basis that is independent of the truth rule. I am now in a position to refine this conclusion, in light of the distinction between the psychological priority of reason and the epistemic priority of reason. Consider the overall argument for the epistemic priority of reason. This argument proceeds, via the infallibility rule, to establish the infallibility of uncorrected reason and hence (a) the asymmetry with respect to infallibility between the uncorrected faculties; it then continues, via (a–c), to establish (d) the epistemic priority of reason. How is the italicized claim in the preceding paragraph to be construed, in the context of its contribution to the argument for the infallibility of uncorrected reason, and hence to the argument for the epistemic priority of reason? If it is understood in terms of epistemic priority, there is a circularity internal to the argument for the epistemic priority of reason—the epistemic priority of reason is required to establish the infallibility of uncorrected reason, and the asymmetry with respect to infallibility, in the first place. If, on the other hand, the italicized claim is understood in terms of psychological priority, the circularity internal to the argument for the epistemic priority of reason vanishes. In other words, we can take the italicized claim to mean that the proper use of reason does not require tests for subversion by sense-perception. This is (the first half of) the psychological priority of reason, a doctrine which is established with reference to the asymmetry with respect to irresistibility and the objective of doxastic permanence, and without relying on the epistemic priority of reason. The issue addressed in this chapter is not whether we should accept (A–D) rather than (a–d). The chapter does attempt, as I indicated in §1, to locate

78

reflection and the stability of belief

Descartes’ ultimate grounds for the priority of reason. As we have seen, the priority of reason divides into two components, psychological and epistemic. The argument for these components proceeds in three stages. In the first stage, the psychological priority of reason is derived from the asymmetry with respect to irresistibility, together with the objective of doxastic permanence—this is the argument at (A–D). In the second stage, the asymmetry with respect to infallibility between the uncorrected faculties is derived from the psychological priority of reason, via the infallibility rule—this is the argument for (a). In the third stage, the epistemic priority of reason is derived from the asymmetry with respect to infallibility, together with the objective of true belief—this is the argument at (a–d). The psychological asymmetry is therefore fundamental to the priority of reason, whether psychological or epistemic.

6. The Priority of Reason and the Cartesian Circle An important obstacle to the success of the Cartesian argument for the epistemic priority of reason remains. The argument for the asymmetry with respect to infallibility between the uncorrected faculties of reason and sense-perception, and hence for the epistemic priority of reason, is threatened by the problem of the Cartesian circle. The argument for the asymmetry with respect to infallibility relies on the infallibility rule. Descartes takes this rule to be a consequence of the existence of a nondeceiving God. The argument for the existence of a nondeceiving God is introduced in a context where the reliability of clear and distinct perception has been called into question, and yet the premises of the argument are themselves generated by clear and distinct perception. The argument for the existence of the nondeceiving God appears question-begging, for the usual reasons, and this circularity in turn infects the argument for the infallibility rule, for (a) the asymmetry with respect to infallibility, for (b) the epistemic account of correction, and hence for (d) the epistemic priority of reason. The threatened circularity is not the circularity internal to the argument for the epistemic priority of reason that arises if the argument for the asymmetry with respect to infallibility itself relies on the epistemic priority of reason. This internal circularity has been removed by supposing that the argument for the asymmetry with respect to infallibility relies on the psychological priority of reason. The circularity in question now is external to the argument for the epistemic priority of reason; the difficulty is that the infallibility rule required to establish the asymmetry with respect to infallibility itself depends on the proof of the existence of a nondeceiving God, a proof that appears to beg the question against the skeptical hypothesis of paragraph four of Meditation III. I believe there is a potential solution to the problem of the Cartesian circle that also coheres with my account of the priority of reason. The solution I have in mind has been proposed, apparently independently, by a number

the priority of reason in descartes

79

of commentators.43 The fundamental idea behind the proposal is this. For Descartes, a deduction or demonstrative argument consists of a connected chain or sequence of individual clear and distinct perceptions (cf. Rules III, VII, XI). Descartes holds that individual clear and distinct perceptions are psychologically irresistible. He also holds that the conclusion of a continuous demonstration is itself irresistible at the time one reaches the conclusion, provided one has attended to the entire chain of clear and distinct perceptions that comprise the demonstration (cf. §3). Suppose one has attended to the continuous demonstration (via the metaphysical principles about causation introduced in Meditation III, the claim that deception is an imperfection, et cetera) of the existence of a nondeceiving God, and of the truth rule. It is a consequence of what Descartes takes to be the psychological characteristics of clear and distinct perception that one will irresistibly believe that a nondeceiving God exists, and that the truth rule is true, at the time one reaches those conclusions. On the present proposal, Descartes’ claim for the central line of argument in Meditations III and IV is not that it constitutes a non-question-begging proof of the existence of a nondeceiving God or of the truth rule. His claim, rather, is that attention to the continuous demonstration induces the psychological state of irresistibly believing that a nondeceiving God exists and that the truth rule is true. The outcome of the argument is that one is in the psychological state of irresistibly believing a proposition that is incompatible with the existence of a deceiver who causes or allows one to have false beliefs based on clear and distinct perception. Let us call this sort of proposal a psychological response to the problem of the circle. Why is it important to Descartes to show that the relevant psychological states can be induced? The answer, I suggest, is that permanence is a doxastic objective; the hypothesis of the powerful deceiver introduced at paragraph four of Meditation III is destabilizing, and hence subversive of permanence in belief.44 Care is needed in explaining the sense in which the hypothesis of the powerful deceiver is “destabilizing.” Suppose that at time t one either has an individual clear and distinct perception that some proposition p is true or one believes that some proposition p is true on the basis of a demonstration to which one has continuously attended. Let us say that in either case the belief, at t, that p is true is a current clear and distinct perception. Since Descartes holds that current clear and distinct perceptions are irresistible, the strength of an inclination to believe p on the basis of a current clear and distinct perception cannot be undermined in the presence of the deceiver hypothesis (cf. HR 1.158–59: AT 7.35–36). By contrast, suppose that at time t one is not having a current clear and distinct perception that p, but that one recollects that p once was a current clear and distinct

43. See especially Rubin 1977 and Larmore 1984. 44. This suggestion is similar in spirit to the material cited in note 7, but clearly supplements the versions of the psychological interpretation presented in Rubin 1977 and Larmore 1984.

80

reflection and the stability of belief

perception. Let us say that in this case the belief, at t, that p is true is a recollected clear and distinct perception. Recollected clear and distinct perceptions are not irresistible. The strength of an inclination to believe a recollected clear and distinct perception will be undermined in the presence of the deceiver hypothesis (cf. HR 1.183–85, 224, 2.42–43: AT 7.69–71, 146, 8A.9–10).45 The deceiver hypothesis is, in this sense, destabilizing with respect to recollected clear and distinct perceptions. This result applies to any belief generated by reason at those times when the beliefs are recollected rather than current.46 From the perspective of the psychological interpretation, the point of attending to the argument of Meditations III and IV is to eliminate the destabilizing effects of the deceiver hypothesis. This is achieved by inducing the psychological state of irresistibly believing that a powerful deceiver does not exist and that the truth rule is true. The psychological interpretation of the procedure in Meditations III and IV extends quite naturally to such conclusions as those at (a–d). The infallibility rule is a consequence of the existence of a nondeceiving God. The infallibility rule is, in turn, used to establish (a) the asymmetry with respect to infallibility, (b) the epistemic nature of correction, and (d) the epistemic priority of reason. Suppose that one has attended to the continuous demonstration (via the existence of a nondeceiving God) of the infallibility rule and of (a), (b), and (d). It is a consequence of what Descartes takes to be the psychological characteristics of clear and distinct perception that one will irresistibly believe the infallibility rule, and (a), (b), and (d), at the time one reaches those conclusions. Thus, one will irresistibly believe that the infallibility rule is true, that there is an asymmetry with respect to infallibility between uncorrected reason and uncorrected sense-perception, that correction is an epistemic process, and so forth. On this account, Descartes’ claim for the arguments for these conclusions is not that they constitute a nonquestion-begging proof of the infallibility rule, (a) the asymmetry with respect to infallibility, (b) the epistemic nature of correction, and (d) the epistemic priority of reason. His claim, rather, is that continuous attention to the demonstrations induces the psychological state of irresistibly believing

45. I think Descartes’ point is not simply that, in the presence of the deceiver hypothesis, our degree of confidence in a recollected conclusion of reason is diminished; I think his point is that we are left without any determinate degree of confidence in the recollected conclusion. 46. There is an additional case in which the presence of the deceiver hypothesis is destabilizing. Consider beliefs based on sense-perception that are sustained by reason—case (iii) in §3. Suppose that at time t one believes, on the basis of a demonstration to which one has continuously attended, that some proposition p generated by sense-perception survives all possible tests for correction. Let us say that the belief, at t, that p is true is a current sustained belief. The belief that p, the current sustained belief, is not itself clearly and distinctly perceived and is therefore not itself irresistible. (It is the belief that p survives all possible tests for correction that is a current conclusion of reason.) The strength of this less than irresistible inclination to believe that p is true will presumably be subverted in the presence of the deceiver hypothesis. Thus, whereas the deceiver hypothesis is destabilizing with respect to only those clear and distinct perceptions that are recollected, it is destabilizing even to current sustained beliefs.

the priority of reason in descartes

81

the propositions in question. If the psychological response can meet the threat of circularity in Descartes’ argument for the existence of a nondeceiving God and for the truth rule, it can equally meet the threat of circularity in his argument for the infallibility rule (itself a premise for the derivation of the truth rule), the asymmetry with respect to infallibility, and the epistemic priority of reason. If the psychological response can be sustained, the problem of circularity is removed as an obstacle to the success of the argument for these results.47 I am not, however, endorsing the psychological response to the problem of the Cartesian circle, either as the best interpretation of what Descartes takes himself to be doing in Meditations III and IV or as a philosophically satisfactory solution to the problem of the circle. One could accept my psychological account of the priority of reason without accepting the psychological response to the problem of the circle. My exposition of the psychological response, together with its extension to (a), (b), and (d), does not include any general assessment of the strengths and weaknesses of this proposal.48 I do offer one observation that is germane to assessing both the psychological interpretation of Descartes’ argument in Meditations III and IV and my psychological interpretation of the priority of reason. These interpretations rely on what are fundamentally the same interpretive resources. In §3, I have argued that the priority of reason is grounded in an asymmetry with respect to irresistibility, together with the objective of doxastic permanence. In §5, I have shown that Descartes can exploit the psychological priority of reason in order to construct an argument for the epistemic priority of reason. This argument is threatened by the problem of the Cartesian circle. In §6, I have noted that on the psychological response, Descartes’ claim for this argument is that, in the interest of doxastic permanence, it induces the psychological state of irresistibly believing such conclusions as the infallibility rule, (a), (b), and (d). The psychological interpretations of the priority of reason, and of Descartes’ argument in Meditations III and IV, both rely on

47. Other obstacles remain. The argument for (a) relies on the infallibility rule, and the infallibility rule is inferred from the existence of a nondeceiving God. Thus, just as the argument for (a) inherits the problem of the Cartesian circle, it also inherits the substantive defects in the arguments for the existence of God in Meditation III. 48. I do believe that the psychological response is by far the most promising constructive attempt to meet the threat of circularity in Meditations III and IV. I suspect that if this interpretation cannot be sustained, dissimulation hypotheses—as discussed in my 1986 [this volume, ch. 1] and 1988—will merit yet increased consideration. What would remain of Cartesian epistemology in the event that Descartes was not sincere in offering his proofs of the existence of a nondeceiving God, and his arguments for the infallibility rule and truth rule, in Meditations III and IV? The present chapter contains an answer: the mandate or sanction, constructed in §3, for holding beliefs generated by the proper use of the cognitive faculties remains intact. Descartes’ argument for (D) the psychological priority of reason, unlike this argument for (d) the epistemic priority of reason, does not appeal to the existence of a nondeceiving God.

82

reflection and the stability of belief

Descartes’ doctrines about the psychological features of the cognitive faculties, and both rely on Descartes’ doctrine that permanence in belief is a doxastic objective. Jointly, these psychological interpretations offer an integrated account of fundamental features of Descartes’ epistemological position. Those who find the psychological response to the problem of the circle attractive ought to welcome my psychological account of the priority of reason.

7. Permanence and Irresistibility in Descartes and Hume Fundamental to the epistemology of Hume is a distinction between two kinds of principles: But here it may be objected, that the imagination, according to my own confession, being the ultimate judge of all systems of philosophy, I am unjust in blaming the antient philosophers for makeing use of that faculty, and allowing themselves to be entirely guided by it in their reasonings. In order to justify myself, I must distinguish in the imagination betwixt the principles which are permanent, irresistable, and universal; such as the customary transition from causes to effects, and from effects to causes: And the principles, which are changeable, weak, and irregular; such as those I have just now taken notice of. The former are the foundation of all our thoughts and actions. . . . The latter are neither unavoidable to mankind, nor necessary, or so much as useful in the conduct of life; but on the contrary are observ’d only to take place in weak minds, and being opposite to the other principles of custom and reasoning, may easily be subverted by a due contrast and opposition. For this reason the former are received by philosophy, and the latter rejected. (T 225) Hume is here classifying principles of the imagination, that is, psychological belief-forming mechanisms. I will refer to the principles that are permanent and irresistible as “PI-principles,” and to the principles that are changeable and weak as “non-PI-principles.” (The contrast between universal and nonuniversal, or “irregular,” principles does not require discussion for the purposes of this chapter.) I will say derivatively that a belief is permanent and irresistible if it is produced by a PI-principle, and that a belief is changeable and weak if it is produced by a non-PI-principle. We can thus distinguish between “PI-beliefs” and “non-PI-beliefs.” When Hume writes of the nonPI-principles as “being opposite” to the PI-principles, I take him to mean that the two sets of principles generate conflicting beliefs. When he writes that the non-PI-principles “may easily be subverted,” I take him to mean that non-PI-beliefs are subverted by the PI-beliefs with which they conflict. When he writes that “philosophy receives” the PI-principles, I take him to

the priority of reason in descartes

83

mean that philosophy receives both PI-principles and the PI-beliefs they generate. We can think of the two sets of psychological principles as constituting two respective faculties, the PI-faculty and the non-PI-faculty.49 One point in the passage cited is that because the PI-principles are irresistible, whereas the non-PI-principles are weak, beliefs generated by the non-PI-faculty “may easily be subverted” or undermined by beliefs generated by the PI-faculty, but not vice versa. In other words, whereas beliefs based on the non-PIfaculty are susceptible to correction by the PI-faculty, beliefs based on the PI-faculty are not susceptible to correction by the non-PI-faculty—the PI-faculty and the non-PI-faculty are hierarchically ordered. “Correction” in this context is, of course, construed psychologically (see §§3 and 5). In addition to observing that non-PI-beliefs are subverted by the PIbeliefs with which they conflict, Hume claims that it is “for this reason” that the PI-principles “are received by philosophy.” The objection to allowing oneself to be “entirely guided” by the principles of the imagination, to include the non-PI-principles, is that this is a suboptimal strategy for achieving permanence in one’s beliefs. This is confirmed by independent discussions of Hume’s sole example of a PI-principle at page 225, the customary transition from causes to effects. Hume states earlier in the Treatise that it is “by their . . . settled order” that ideas “arising from . . . the relation of cause and effect . . . distinguish themselves from the other ideas, which are merely the offspring of the imagination” (T 108). Hume writes two pages later that one of the “advantages” of beliefs based on the relation of cause and effect is that they are “solid . . . and invariable” (110). Hume’s assertion that PI-beliefs are “received by philosophy” is his way of assigning such beliefs normative pride of place. Permanent and irresistible belief-forming mechanisms lead to a more permanent structure of beliefs than any available cognitive alternative. 50 It is noteworthy that the objective of achieving permanence and avoiding

49. I believe that Hume sometimes identifies the set of non-PI-principles with the imagination (T 117–18n., 371n.), though he also uses this term more broadly to refer to the set of PIprinciples and the set of non-PI-principles, taken together (225, 267). I introduce the terms ‘PI-faculty’ and ‘non-PI-faculty’ in order to avoid, insofar as possible, interpretation of Hume’s terminology. My substantive argument that follows, however, does require the claim that Hume identifies the set of PI-principles with the understanding. 50. Previous interpretations of Hume’s rationale for approving of the PI-principles place insufficient emphasis on the underlying importance of permanence in belief. Lenz maintains that Hume justifies causal inference on the ground that it is universal (1958, 564) and irresistible (566). Lenz does notice that only beliefs based on PI-principles will be permanent. However, he treats this not as the basis of their being received by philosophy, but as an incidental by-product of their irresistibility (567). Passmore holds that Hume takes the PI-principles to be “natural” because they promote “consistent and orderly thinking” or “regularity” (1952/1968, 55). Passmore’s interpretation is similar to mine in spirit, but he neglects, or at least does not make explicit, the temporal aspects of consistency and orderliness.

84

reflection and the stability of belief

fluctuation or contradiction also plays a crucial role in Hume’s account of moral judgment at Selby-Bigge pages 581–82 of the Treatise and 227– 29 of the second Enquiry.51 Furthermore, the non-PI-faculty itself generates conflicting beliefs. This point emerges in the paragraph following the initial characterization of the distinction between the PI-principles and non-PI-principles: The opinions of the antient philosophers . . . are deriv’d from principles, which, however common, are neither universal nor unavoidable in human nature. The modern philosophy pretends to be entirely free from this defect, and to arise only from the solid, permanent, and consistent principles of the imagination. (T 226) The non-PI-principles are evidently inconsistent, in the sense that they generate conflicting beliefs (that is, inclinations to hold incompatible beliefs). This is confirmed in the conclusion of Book I. In discussing the option of assenting “to every trivial suggestion of the fancy,” Hume writes that “these suggestions are often contrary to each other” (267). In the following paragraph, Hume contrasts “the trivial suggestions of the fancy” with “the understanding, that is, . . . the general and more establish’d properties of the imagination” (267). These “general and more establish’d properties” are the PI-principles of pages 225–26; the trivial suggestions of the fancy are identical to the non-PI-beliefs.52 If the trivial suggestions of the fancy are inconsistent, the non-PI-principles are inconsistent.53 Hume holds not only that the non-PI-faculty generates beliefs that conflict with beliefs generated by the understanding but also that the non-PI-faculty itself generates conflicting beliefs. There is an impressive structural similarity between Hume’s position and that of Descartes. For Descartes, reliance on sense-perception on its own leads to ineliminable incoherence; reason, in virtue of its irresistibility, corrects sense-perception by suppressing some of the inclinations to believe that sense-perception generates. For Hume, reliance on the nonPI-faculty on its own leads to ineliminable incoherence; the PI-faculty, in virtue of its irresistibility, corrects the non-PI-faculty by suppressing some

51. For discussion of the role that Hume assigns to permanence in moral judgments, see Atkinson 1976, esp. pt. IV; Baier 1982; Hearn 1976, esp. pt. I; Jones 1976, esp. 328–30; Mackie 1980, esp. ch. VII; and MacNabb 1951/1966, 191–97. 52. There is additional evidence for this claim. The initial distinction between the PI-principles and non-PI-principles is introduced in the first two paragraphs of I.iv.4., “Of the modern philosophy.” At page 225, when Hume mentions as examples of non-PI-principles “those I have just now taken notice of,” he is referring to the psychological mechanisms discussed in the preceding section, “Of the antient philosophy.” In the final paragraph of that section, Hume writes that the ancient philosophers “were guided by every trivial propensity of the imagination” (T 224). 53. The role Hume assigns to the trivial propensities of the imagination receives insufficient attention. Immerwahr is a notable exception. See his 1977 and 1979.

the priority of reason in descartes

85

of the inclinations to believe that the non-PI-faculty generates.54 (The language of “inclination” and “suppression” is Hume’s—see T 224.) In sum, the PI-faculty stands in the same relationship to the non-PI-faculty for Hume as reason stands to sense-perception for Descartes. Descartes and Hume also provide substantially the same rationale for submitting beliefs generated by one faculty to tests for subversion by another faculty. Cognitive faculties are evaluated with reference to their contribution to doxastic permanence.55 One important determinant of that contribution is the relative strength of the inclinations to hold beliefs generated by the available faculties or belief-forming mechanisms; greater degrees of irresistibility contribute to greater degrees of permanence. Given the psychological properties of the relevant faculties, submitting the psychologically weaker faculty to tests for correction by the stronger faculty generates a more permanent belief system than any available cognitive alternative. Descartes and Hume share a common project, to show how permanence can be achieved. It may not be surprising that we find in Plato passages that have affinities with Descartes’ view that there are conflicts within sense-perception that must be resolved by reason.56 What is striking is the presence of the similarities in structure and underlying rationale that I have identified in the Rationalist epistemology of Descartes and the Empiricist epistemology of Hume. Both hold that the cognitive faculties are hierarchically ordered in virtue of their psychological properties. It is not my intent to obscure the differences between the epistemological theories of Descartes and Hume. The most obvious difference in the theories, as developed to this point, is in the locus of irresistible beliefs. For Descartes, it is only clear and distinct perception that is psychologically compelling; for Hume, the psychologically compelling beliefs include those that result from “the customary transition from causes to effects,” from associative mechanisms triggered by a background of

54. Descartes and Hume would disagree about how to characterize the notion of an “incoherence” (or “conflict”) that arises within a faculty, or between faculties. Descartes would characterize this notion in terms of logical incompatibility. Hume, I believe, would characterize it in terms of psychological instability—a faculty or set of faculties is incoherent to the extent that it generates psychologically unstable sets of doxastic states (that is, unstable sets of beliefs and dispositions to hold beliefs). On this model, dispositions to hold logically incompatible beliefs would be a special case of incoherence. The notion of doxastic instability, implicit in Descartes, is a major preoccupation for Hume, who devotes many pages to the psychological dynamics of the imagination’s attempts to resolve such conflicts (cf. T 186–87, 198–210, 210–16, 219–24, 253–55). Space does not permit development of this point. 55. In this regard, both figures have an affinity with Charles S. Pierce, who in “The Fixation of Belief” takes “the settlement of opinion” or “firm belief “ as “the sole end of inquiry” (CP 5.375). In “How to Make Our Ideas Clear,” Pierce writes of Leibniz that “He . . . missed the most essential point of the Cartesian philosophy, which is, that to accept propositions which seem perfectly evident to us is a thing which, whether it be logical or illogical, we cannot help doing” (CP 5.392). [For ‘CP’, see chapter 4, note 7.] 56. Cf. Republic 523a–524d and Theaetetus 184b–187a. The interpretation of these passages is, of course, controversial. For some discussion, see Cooper 1970 and Holland 1973.

86

reflection and the stability of belief

observed conjunctions.57 This difference is an important symptom, if not the essence, of the difference between the epistemological orientations of Descartes and Hume. There is one additional development in Hume’s theory that must be mentioned. Hume brings himself to conclude that there is internal incoherence within the PI-faculty itself: Thus there is a direct and total opposition betwixt our reason and our senses; or more properly speaking, betwixt those conclusions we form from cause and effect, and those that persuade us of the continu’d and independent existence of body. (T 231) ’Tis this principle [the imagination], which makes us reason from causes and effects; and ’tis the same principle, which convinces us of the continu’d existence of external objects, when absent from the senses. But tho’ these two operations be equally natural and necessary in the human mind, yet in some circumstances they are directly contrary. (266) Two “equally natural and necessary” principles of the imagination, in other words, the PI-principles themselves, generate conflicting beliefs.58 (It should be noted that Hume is not basing his conclusion on such conflicts as arise from perceptual relativity. He thinks the latter conflicts are resolved within the PI-faculty through the use of “general rules”—cf. 632/1.3.10.12.) Since PI-principles generate beliefs that are irresistible, this conflict, unlike those generated within the non-PI-faculty, cannot be resolved.59 The continuation of the passage at page 266 of the Treatise explicitly draws out the consequence that permanence cannot be achieved: [N]or is it possible for us to reason justly and regularly from causes and effects, and at the same time believe the continu’d existence of matter. How then shall we adjust those principles together? Which of them shall we prefer? Or in case we prefer neither of them, but successively assent to both, as is usual among philosophers, with what confidence can we afterwards usurp that glorious title, when we thus knowingly embrace a manifest contradiction?

57. This is Hume’s example at Treatise 225; however, for Hume there must be other psychological propensities that generate irresistible beliefs. This is a consequence of his discussions at Treatise 231 and 266, quoted later in this section. 58. For discussions of Hume’s claim, see Bricke 1980, 9–12; Butler 1960, 80–81, 85; Kemp Smith 1941, 127–29 (cf. 490–94); and Robison 1976, 46–48. The argument at Treatise 226–31 that leads Hume to the claim in question is not relevant to the present chapter. For a careful exposition of the argument, see Bricke 1980, 16–19. 59. For a different interpretation of the outcome, see Bricke 1980, 19–24. A full discussion of the issues is not possible here.

the priority of reason in descartes

87

The conclusion, if correct, that the PI-faculty generates ineliminable incoherence undermines the entire project of justifying beliefs with reference to the objective of doxastic permanence. Descartes is confident that the source of irresistible beliefs is coherent (cf. §2). This confidence stands in sharp contrast to Hume’s despairing conclusion that the source of irresistible beliefs is incoherent.60

60. I am indebted to Nicholas White for urging me to write a paper on the present topic, and to Jaegwon Kim for questions concerning my “Is There Radical Dissimulation in Descartes’ Meditations?” [this volume, ch. 1] that led me to develop the interpretive material at §§2–3. I am also grateful to Richard Brandt, Tyler Burge, Stephen Darwall, Jaegwon Kim, Brian McLaughlin, Adrian Piper, Frederick Schmitt, Lawrence Sklar, and Nicholas White, for helpful comments. I have benefited from reading versions of this chapter at Bowling Green State University and Southern Methodist University. The chapter has been improved in response to the comments of anonymous referees and editors at The Philosophical Review. I am especially grateful to David Velleman for providing detailed substantive, organizational, and stylistic comments on previous versions.

88

reflection and the stability of belief

3 The Cartesian Circle

1. The Truth Rule and the Problem of the Cartesian Circle descartes writes in the second paragraph of Meditation III: “So I now seem to be able to lay it down as a general rule that whatever I perceive very clearly and distinctly is true” (CSM 2.24: AT 7.35).1 I call this principle the truth rule. In the third paragraph, Descartes decides that it is premature to take the truth rule to be established. He writes of “very simple and straightforward” propositions in arithmetic and geometry: “the . . . reason for my . . . judgment that they were open to doubt was that it occurred to me that perhaps some God could have given me a nature such that I was deceived even in matters which seemed most evident” (CSM 2.25: AT 7.36). The matters that seem most evident, in the context of paragraph two, are beliefs based on clear and distinct perception, so that these beliefs (together with any that seem less evident) are themselves open to doubt. Descartes writes: “in order to remove . . . this . . . reason for doubt, . . . I must examine whether there is a God, and, if there is, whether he can be a deceiver” (CSM 2.25: AT 7.36). In Meditation III, Descartes offers an argument for the existence of a nondeceiving God. The truth rule is finally proved in Meditation IV. Descartes concludes, on the ground that God is no deceiver, that “if . . . I restrain my will so that it extends to what the intellect clearly and distinctly reveals, and no further, then it is quite impossible for me to go wrong” (CSM 2.43: AT 7.62). 1. [“Abbreviations for Editions of Seventeenth- to Nineteenth-Century Works” contains the full reference system for this chapter. Section numbers have been added in the text.]

89

Descartes’ procedure has been thought to suffer from an obvious difficulty. The truth rule is proved after even the most evident beliefs have been placed in doubt. The premises of the argument for the truth rule in Meditations III and IV can at best be matters that seem most evident, matters that are themselves open to doubt in light of the supposition of a deceiving God. Even if the premises for the demonstration of the truth rule are confined to beliefs based on clear and distinct perception, Descartes’ argument relies on premises whose truth has been called into question in order to show that he is not deceived in these very matters. The problem of “the Cartesian circle” is the problem of acquitting Descartes of the charge that his procedure is question-begging. An enormous literature offers a bewildering variety of solutions to this problem. I believe that two broad lines of interpretation now vie with each other as solutions.2 We can highlight the difference between them with reference to Descartes’ claim that he must consider whether there exists a deceiving God in order to “remove” (tollere) the reason for doubt. According to the first interpretation, Descartes holds that the argument for the truth rule removes the reason for doubt in that it provides a good reason not to doubt beliefs based on clear and distinct perception, or at least shows that there is no good reason to doubt them. I call this the epistemic interpretation. Proponents of this interpretation include Curley, Doney, Frankfurt, and Gewirth.3 According to the second interpretation, Descartes holds that the argument for the truth rule removes the reason for doubt in that it renders it psychologically impossible to doubt beliefs based on clear and distinct perception, or at least enables one to attain a state in which it is psychologically impossible to doubt them. I call this the psychological interpretation. Larmore and Rubin are most clearly proponents of this interpretation.4 Bennett is a proponent of a guarded version of the interpretation.5 I believe 2. Interpretations in which memory, not clear and distinct perception, is called into doubt have not been a live option in the aftermath of the critique of Frankfurt 1962 and 1970, ch. 14. For the obituaries, see Doney 1970, 671n.4; Prendergast 1972, 304; Sanford 1973, 122; Curley 1978, 104; B. Williams 1978, 193n.7; and Van Cleve 1979, 56–57. There was already substantial criticism of the memory interpretation in Merrylees 1934, ch. 4, §§3–4; Levett 1937; Laporte 1950, 162–63; and Wolz 1950, 481–82. For additional criticism that is independent of Frankfurt, see Gouhier 1962, 302–5; Beck 1965, 147–48; and Etchemendy 1981, §§2–3. 3. See Gewirth, 1941, 1970, and 1971; Frankfurt 1965 and 1970, esp. ch. 5; Doney 1970; and Curley 1978, ch. 5. Doney 1970 abandons the memory interpretation he had espoused in 1955. 4. See Rubin 1977 and Larmore 1984. Indeed, Cottingham (1986, 69, 76–77) takes Frankfurt 1970 to be an instance of the psychological interpretation, and Markie (1986, 43–44, to include n.4) cites Curley 1978 and Gewirth 1941 as providing a psychological account of how the reason for doubt is removed. The interpretation and classification of positions in the literature is itself a difficult matter. 5. Bennett takes the psychological response to the problem of the circle, and related doctrines, to represent a “lesser strand” that does not constitute Descartes’ “principal, official account” of his procedure, though it is in Descartes’ mind “at some level” (1990, §§1, 10, 12, 14). This essay was nearing completion when Bennett’s (then unpublished) manuscript came to my attention.

90

reflection and the stability of belief

the psychological interpretation merits a more sustained development than it has received. This interpretation has not crystallized in the literature, even though a good deal of recent work on the problem of the circle points in its direction. In this chapter, I explore the textual merit of what I take to be the most promising version of the psychological interpretation.

2. A Psychological Interpretation of Unshakability and Scientific Knowledge In a 1640 letter to Regius, Descartes writes of a “conviction based on an argument so strong that it can never be shaken by any stronger argument” (K 74: AT 3.65). A belief could be shakable before one comes to possess a particular argument, and unshakable thereafter (cf. CSM 2.309: AT 7.460). A person’s belief is unshakable precisely when the person possesses arguments that prevent the belief from being shaken by argument. How is this condition to be understood? Descartes frequently writes of beliefs that are firm or solid (CSM 1.126, 2.12, 103, 104: AT 6.31, 7.17, 145, 146). Firmness is explicitly associated with unshakability in the Second Replies, where Descartes writes of “a conviction so firm that it is quite incapable of being destroyed (tollere)” (CSM 2.103: AT 7.145). The metaphor is also associated with unshakability in The Search after Truth, a work that contains persistent references to the notion of a firm or solid basis for knowledge (CSM 2.400, 405, 407, 408: AT 10.496, 506, 509, 513). Beliefs are not firm if arguments can “overturn” (renverser) them (CSM 2.408: AT 10.512, 513). These passages suggest that a belief is unshakable just in case the person possesses arguments that prevent the belief from being dislodged by argument.6 I say for brevity that an unshakable belief cannot be dislodged by argument, or cannot be dislodged. Descartes’ characterization of unshakability, in the passages cited from the letter to Regius and the Search, is not epistemic. These passages do not say that an unshakable belief is one that it would be unreasonable, or unjustified, or unwarranted to disturb or relinquish in the face of argument.7

6. Bennett’s “stability” is a generalized notion of unshakable belief, belief that cannot be dislodged by argument or otherwise (cf. 1990, §4). 7. The fact that “unshakability” is characterized with reference to the notion of an “argument” [as at K 74] or “reason” [as in the modification of the Kenny translation at CSMK 147] (ratio: AT 3.65; cf. 7.69, 70) does not render the account “epistemic,” rather than “psychological,” in my senses of these terms. Suppose, for the sake of discussion, that the notion of an “argument” can only be characterized epistemically. The unshakability of a belief is relative to the arguments one possesses; unshakability is a relational property. We can formulate epistemic accounts of this relation, e.g., a person’s belief is unshakable just in case the person possesses arguments that constitute good reasons for not relinquishing the belief in the face of argument. We can formulate psychological accounts of this relation, e.g., a person’s belief is unshakable just in case the person possesses arguments that, as a psychological matter of fact, prevent the

the cartesian circle

91

More generally, these passages do not provide a normative characterization of “unshakability.” They do not say that an unshakable belief is one that one ought not disturb or relinquish in the face of argument. An unshakable belief is one that cannot be dislodged by argument. Whether a belief is unshakable is a question of descriptive psychology. This is not to deny that we can locate discussions of unshakability that seem more epistemic or normative in character. The important point is that a psychological account of unshakability is available for our use.8 Although unshakability is not itself a normative notion, Descartes regards unshakability or firmness as a doxastic objective—a goal which our beliefs ought to attain. This is implicit in the discussion in the Search of how to achieve firmness (cf. CSM 2.406–9: AT 10.509–13). It is explicit in the first paragraph of Meditation I, where Descartes writes of “the necessity to start again right from the foundations to establish anything at all in the sciences that was stable (firmum)” (CSM 2.12: AT 7.17), and in the first paragraph of the Search, where he formulates the objective of laying “the foundations for a solid science” (CSM 2.400: AT 10.496). The metaphor of a firm foundation also appears in Parts II and IV of the Discourse on the Method (CSM 1.117–18, 126: AT 6.12–14, 31) and is developed at length in the Seventh Replies (CSM 2.365–80: AT 7.536–56).9 Descartes writes in the Second Replies: First of all, as soon as we think that we correctly perceive something, we are spontaneously convinced that it is true. Now if this conviction is so firm that it is impossible for us ever to have any reason for doubting what we are convinced of, then there are no further questions for us to ask: we have everything that we could reasonably want. . . . For the supposition which we are making here is of a conviction so firm that it is quite incapable of being destroyed; and such a conviction is clearly the same as the most perfect certainty. (CSM 2.103: AT 7.144–45)

dislodging of the belief in the face of argument. Of the two italicized accounts of the relation, only the first invokes epistemic notions. We can distinguish epistemic and psychological accounts of unshakability, even if Descartes provides an epistemic as well as a psychological account, and even if he holds that a belief is unshakable in some specified psychological sense just in case it is unshakable in some specified epistemic sense. 8. Tlumak provides a number of epistemic characterizations of unshakability— “irrevisability” in his terminology (1978, 45, 46, 48). He explicates “metaphysical certainty” as irrevisability (44–45) and holds that metaphysical certainty “has normative force in the context of the Cartesian problem” (43). I believe the normativity of the notion derives from Descartes’ adoption of unshakability as an objective of inquiry (see §2). We need not regard the normative characterization as fundamental. This point undermines the second of Tlumak’s arguments against the psychological interpretation (57). 9. Bennett details the way in which this material supports the psychological interpretation (1990, §1).

92

reflection and the stability of belief

This is perhaps Descartes’ most developed statement of unshakability as an objective of inquiry.10 The Second Replies identifies unshakable belief with “perfect certainty.” The letter to Regius identifies unshakable belief with scientia, “scientific knowledge” (AT 3.65). There is related terminology elsewhere: “certain science” (CSM 1.197: AT 8A.10), “true knowledge” (CSM 2.101: AT 7.141), “true and certain knowledge” (CSM 2.48: AT 7.69), “perfect knowledge” (CSM 2.49: AT 7.71), and a proposition’s being “perfectly known” (CSM 2.48: AT 7.69). Such expressions appear to be terminological variants of the notion of scientific knowledge.11 Knowledge, in the strict sense of scientific knowledge, is identified with unshakable belief, and hence itself has a psychological characterization.12

3. The Skeptical Supposition, Shakability, and Doubt In explaining the reason for doubt in Meditation III, Descartes does not claim that he has any reason to believe that a deceiving God exists; rather, “it occurred to me that perhaps” God is a deceiver. He points out that he has “no cause to think that there is a deceiving God” and that he does “not yet even know for sure whether there is a God at all” (CSM 2.25: AT 7.36). Descartes appeals to these features of his situation in observing that “any reason for doubt which depends simply on this supposition [of a deceiving God] is a very slight and, so to speak, metaphysical one” (CSM 2.25: AT 7.36). To suppose that God is a deceiver need not be to believe that God is a deceiver. I suggest that we understand the notion of supposing broadly; to suppose that p is to believe, assume, hypothesize, conjecture, suspect, conceive, or imagine that p. To suppose that p is to be in the psychological state of holding one of these propositional attitudes toward p. The supposition that there exists a deceiving God is not, strictly speaking, the only supposition that renders even the most evident beliefs doubtful:

10. The reference to a “reason for doubting” suggests an epistemic reading of the operative notion of unshakability. As Bennett observes, however, the Latin causa can also mean “cause,” a rendering that is more consistent with the remainder of the passage (1990, 253n.15). 11. For discussions of one or both of the interrelated concepts of unshakability and scientific knowledge, see Doney 1955, 337; Kenny 1968, 191–93; Doney 1970, esp. 388–91; Frankfurt 1970, 24, 124, 179–80; Curley 1978, 104–5; Tlumak 1978, 44–50, 57–60; B. Williams 1978, 62, 200–204; Gombay 1979, 491–94; B. Williams 1983, 345; Cottingham 1986, 2–3, 25, 67, 70–71; Markie 1986, 59–72; Rodis-Lewis 1986, 277–81; and Garns 1988, esp. §§1–2. 12. Although he holds that Descartes employs epistemic and psychological conceptions of certainty, Markie (1986, 34, 53–57) offers only an epistemic account of scientific knowledge (59–72). This leaves Markie, on his own admission, without a solution to the problem of the circle (cf. 162n.12).

the cartesian circle

93

As for the kind of knowledge possessed by the atheist, it is easy to demonstrate that it is not . . . certain. As I have stated previously, the less power the atheist attributes to the author of his being, the more reason he will have to suspect that his nature may be so imperfect as to allow him to be deceived even in matters which seem utterly evident to him. (CSM 2.289: AT 7.428) The supposition that one is caused by something less powerful than God is itself a cause for doubt. The supposition, in its most general form, that renders beliefs based on clear and distinct perception doubtful is that one’s faculty of clear and distinct perception is defective—whether as the result of a deceiving God, a powerful demon, some other chain of events, or chance (cf. CSM 2.14: AT 7.21). I refer to the propositional content indicated in italics as the skeptical hypothesis. If the truth rule is true—if whatever one clearly and distinctly perceives is true—then the skeptical hypothesis is false. I refer to the supposition that the skeptical hypothesis is true as the skeptical supposition. Descartes claims in paragraph fourteen of Meditation V that beliefs based on clear and distinct perception are shakable insofar as they are held by someone who lacks knowledge of (the existence of a nondeceiving) God and of the truth rule: And so other arguments can now occur to me which might easily undermine my opinion, if I did not possess knowledge of God; and I should thus never have true and certain knowledge about anything, but only shifting and changeable opinions. . . . I can easily fall into doubt . . . , if I am without knowledge of God. For I can convince myself that I have a natural disposition to go wrong from time to time in matters which I think I perceive as evidently as can be. (CSM 2.48: AT 7.69–70) Knowledge of the truth rule is a necessary condition for scientific knowledge. (I defer the question of what constitutes knowledge of the truth rule in this context.) The psychological account of unshakability is operative here. A person who lacks knowledge of the truth rule can suppose that clear and distinct perception is defective. Descartes does not state that, in light of this supposition, one ought to disturb or relinquish beliefs based on clear and distinct perception; rather, the supposition can “undermine” (deicere) or dislodge those beliefs. Furthermore, someone who supposes that the skeptical hypothesis is true thereby falls into doubt. This point, which occurs in a number of additional passages (cf. CSM 1.197, 2.101, 289: AT 8A.9–10, 7.141, 428), establishes a connection between a belief’s being doubtful and a belief’s being shakable. The fact that someone who lacks knowledge of the truth rule can suppose that the skeptical hypothesis is true, and thereby fall into doubt, would not show that the person’s beliefs are shakable, unless doubt is a state that has

94

reflection and the stability of belief

the psychological property of being able to dislodge belief. This connection is confirmed in the Second Replies. Descartes writes: “I maintain that this awareness of his is not true knowledge, since no act of awareness that can be rendered doubtful seems fit to be called knowledge” (CSM 2.101: AT 7.141). Four (Adam and Tannery) pages later, he identifies “perfect certainty” with “a conviction so firm that it is quite incapable of being destroyed” (CSM 2.103: AT 7.145). Belief that is doubtful is not scientific knowledge. Doubt must therefore be a state that has the psychological property of being able to dislodge belief. I suggest that doubt is able to dislodge belief in virtue of its unsettling or destabilizing belief. A belief that is unstable is liable to be dislodged, though it might remain in place. This model generates a coherent picture of relevant texts. When a belief is “shaken” (concutere), as in the letter to Regius, it is destabilized.13 The persistent metaphors of firm and solid belief (firmus in Latin and firme or solide in French) are to be understood in terms of stability.14 Unshakable or firm belief is belief that cannot be destabilized by argument. The reference to shifting or inconstant (vagas) belief in the Meditation V passage also suggests that doubt is destabilizing.15 Descartes does not hold that doubt is sufficient to dislodge belief; he holds that doubt (until such time as it is removed) is sufficient to destabilize belief.16 The fact that a belief is destabilized explains how it can be dislodged. It remains the case that a belief is unshakable just in case it cannot be dislodged by argument; an unshakable belief cannot be dislodged by argument because it cannot be destabilized by argument.

4. The Unshakability of Current versus Recollected Clear and Distinct Perceptions The claim that beliefs based on clear and distinct perception are shakable by the skeptical supposition requires qualification. Clear and distinct perception is psychologically compelling in that the belief that p is irresistible at any time p is clearly and distinctly perceived: “my nature is such that so 13. Kenny’s rendering ‘concuti’ as “shaken” (CSMK 147) and Rodis-Lewis’ “ébranlée” (1984, 266) are nearer the mark than Gewirth’s “destroyed” (1941, 390) or B. Williams’ “knocked out” (1978, 204). 14. They are sometimes so translated, for example, at CSM 2.12. See also Etchemendy’s suggested translation of ‘firmanda’ (AT 7.15) in terms of “stabilization” (1981, 37–38). 15. Doney, who frequently invokes the terminology of “unsettled belief” (1970, 390, 391, 401), renders ‘vagas’ as “unstable” (389). 16. The ontologically most economical reading of Descartes’ position would identify doubt with unstable belief. On this reading, the claim that the skeptical supposition renders beliefs based on clear and distinct perception doubtful means that the supposition destabilizes these beliefs. This is a Peircean position: the end of inquiry is settled or firm belief—doubt is a stimulus to this state. See Charles S. Peirce, “The Fixation of Belief,” §§3–4 (CP 5.370–76). (For ‘CP’, see chapter 4, note 7.)

the cartesian circle

95

long as I perceive something very clearly and distinctly I cannot but believe it to be true” (CSM 2.48: AT 7.69). The textual basis for this doctrine is overwhelming (cf. CSM 1.207, 2.27, 103, 309; CSMK 147, 233, 334: AT 8A.21, 7.38, 144, 460, 3.64, 4.115–16, 5.148). Clear and distinct perception divides into intuition, the apprehension of the truth of a proposition all at once or in a moment, and demonstration, a connected sequence of intuitions (cf. Rules for the Direction of the Mind, III, VII, XI). Descartes writes, again in paragraph fourteen of Meditation V: “when I consider the nature of a triangle, it appears most evident to me . . . that its three angles are equal to two right angles; and so long as I attend to the proof, I cannot but believe this to be true” (CSM 2.48: AT 7.69–70). The Principles of Philosophy and the Conversation with Burman reiterate the point that the doctrine of the irresistibility of clear and distinct perception applies to demonstration as well as to intuition (cf. CSM 1.197; CSMK 334–35: AT 8A.9, 30–31, 5.148). The belief that p is psychologically irresistible at any time that one intuits p or attends to a demonstration of p.17 I call the belief that p, at any time one clearly and distinctly perceives p, a current clear and distinct perception. Consider a time at which one is not having a current clear and distinct perception that p, but recollects that one previously clearly and distinctly perceived that p. I call the belief that p, at the time of the recollection, a recollected clear and distinct perception. (Recollected clear and distinct perceptions, as I have characterized them, are merely recollected in the sense that they are not also clearly and distinctly perceived at the time of the recollection.) Beliefs based on clear and distinct perception are either current or recollected clear and distinct perceptions. I call a proposition that a person has intuited an axiom for that person, and a proposition that a person has demonstrated (but not intuited) a theorem. Current clear and distinct perceptions include current axioms and current theorems; recollected clear and distinct perceptions include recollected axioms and recollected theorems.18 Descartes persistently invokes the distinction between current and recollected theorems in passages germane to the circle (see CSM 1.197, 2.48, 100, 104–5, 171; CSMK 147: AT 8A.9–10, 7.69–70, 140, 145–46, 246, 3.64–65). In all but two of these passages (CSM 2.100, 171: AT 7.140, 246), he explicitly maintains that although the belief that p is irresistible so long as it is a current theorem, the belief that p is not irresistible at times it is a recollected theorem. Here is a more extensive quotation from paragraph fourteen of Meditation V:

17. I leave open the question of precisely what is involved in attending to a demonstration. Descartes seems to hold that attending to a demonstration requires that “we are able to grasp the proof . . . in its entirety” (CSMK 335: AT 5.149). 18. It is possible for there to be recollected axioms if it is possible to recollect that one intuited a proposition without intuiting the proposition at the time of the recollection. No Cartesian doctrines preclude this possibility, though some axioms are such that we cannot think of them without clearly and distinctly perceiving them (cf. CSM 2.104: AT 7.145).

96

reflection and the stability of belief

Admittedly my nature is such that so long as I perceive something very clearly and distinctly I cannot but believe it to be true. But . . . often the memory of a previously made judgement may come back, when I am no longer attending to the arguments which led me to make it. And so other arguments can now occur to me which might easily undermine my opinion, if I did not possess knowledge of God; and I should thus never have true and certain knowledge about anything, but only shifting and changeable opinions. For example, when I consider the nature of a triangle, it appears most evident to me . . . that its three angles are equal to two right angles; and so long as I attend to the proof, I cannot but believe this to be true. But as soon as I turn my mind’s eye away from the proof, then in spite of still remembering that I perceived it very clearly, I can easily fall into doubt about its truth, if I am without knowledge of God. For I can convince myself that I have a natural disposition to go wrong from time to time in matters which I think I perceive as evidently as can be. (CSM 2.48: AT 7.69–70) The skeptical supposition cannot dislodge a current theorem, since current theorems are irresistible; it can dislodge recollected theorems (cf. CSM 1.197: AT 8A.9–10). Recollected theorems, evidently, are not psychologically irresistible. More generally, recollected clear and distinct perceptions can be dislodged by the skeptical supposition (cf. CSM 2.309: AT 7.460).19 Recollected clear and distinct perceptions, unlike current clear and distinct perceptions, are not psychologically irresistible (cf. CSMK 353: AT 5.178).20 The psychological doctrine that recollected clear and distinct perceptions can be dislodged by the skeptical supposition will seem more plausible against the background of my suggestion that doubt is a state that destabilizes belief. If this suggestion is correct, we should expect that the skeptical supposition destabilizes belief, since someone who supposes that the skeptical hypothesis is true falls into doubt. The psychological irresistibility of current clear and distinct perceptions is caused by their being clearly and distinctly perceived, not by one’s believing that they are clearly and distinctly perceived. By contrast, one believes a recollected clear and distinct perception, at least in part, on the ground that one previously clearly and distinctly perceived the proposition. Let the belief that p be a recollected clear and distinct perception. Consider the following psychological states: the belief, on the ground that p was clearly and distinctly perceived, that p; and the supposition that the faculty of clear and distinct perception is defective. It seems plausible that these states, taken together, are unstable—especially if 19. Axioms that cannot be thought of without being clearly and distinctly perceived are an exception (cf. note 18). 20. Since doubt is a state that is able to dislodge belief, we should expect Descartes to hold that current clear and distinct perceptions cannot be doubtful. This is Descartes’ position at CSM 2.104, 309, 321; CSMK 353: AT 7.146, 460, 477, 5.178.

the cartesian circle

97

the supposition that clear and distinct perception is “defective” is taken to mean that it is unreliable, that it produces false beliefs more often than true beliefs.21 The skeptical supposition therefore destabilizes the belief that p. This explains how the skeptical supposition can dislodge recollected clear and distinct perceptions. Unshakability could in principle be achieved by avoiding recollected clear and distinct perceptions in favor of current clear and distinct perceptions. A person who followed this policy would reintuit any axiom, and redemonstrate any theorem, at every time he believed the axiom or theorem in question. Because his beliefs based on clear and distinct perception would be confined to current clear and distinct perceptions, and hence be irresistible, they would be unshakable by the skeptical supposition. I believe Descartes would reject this technique for achieving unshakability simply on the ground that humans do not have sufficient conscious cognitive capacity at any time to intuit every axiom, and demonstrate every theorem, that they believe at that time—“the mind cannot think of a large number of things at the same time” (CSMK 335: AT 5.148).22 How can unshakability be achieved? Descartes writes in paragraph fifteen of Meditation V: Now, however, I have perceived that God exists, . . . and I have drawn the conclusion that everything which I clearly and distinctly perceive is of necessity true. Accordingly, . . . there are no counter-arguments which can be adduced to make me doubt it, but on the contrary I have true and certain knowledge of it. And I have knowledge not just of this matter, but of all matters which I remember ever having demonstrated, in geometry and so on. (CSM 2.48: AT 7.70) Knowledge of the truth rule is a sufficient condition, as well as a necessary condition, for the unshakability specifically of recollected clear and distinct

21. Descartes writes, however, of the supposition that clear and distinct perception “from time to time” leads to error (CSM 2.48, 289: AT 7.70, 428), that “the way I am made makes me prone to frequent error” (CSM 2.48: AT 7.70). In Meditation I, by contrast, the skeptical supposition is explicitly that one is deceived “all the time” (CSM 2.14: AT 7.21). (The Haldane and Ross translations of the first and third of these passages obscure the Latin; see HR 1.184 and 1.147, respectively.) For previous discussion of the contrast, see Gouhier 1962, 301–2. It is controversial whether the scope of the skeptical supposition in Meditation I extends beyond sense-perception (cf. Frankfurt 1970, esp. chs. 7–8). Even if it does not, we should expect that if Descartes subjects sense-perception to the supposition that it is systematically defective, he is obliged to subject clear and distinct perception to the analogous supposition. It is surprisingly difficult to find explicit textual evidence that he does so. For additional disanalogies between the skeptical suppositions in Meditations I and III, see my 1986, 247–53 [this volume, ch. 1, §2]. 22. Cf. Etchemendy 1981, 8, and Cottingham 1986, 70, 77n.24. The technique would also sacrifice the objective of achieving a comprehensive system of beliefs (cf. CSM 1.16, 179–80: AT 10.171–72, 9B.2–3).

98

reflection and the stability of belief

perceptions.23 This claim is repeated in the Second Replies (CSM 2.104–5: AT 7.146) and the letter to Regius (CSMK 147: AT 3.65).24 Consider a proposition that one previously clearly and distinctly perceived, and such that one retains belief in the proposition, without either clearly and distinctly perceiving the proposition, or remembering that one clearly and distinctly perceived the proposition. The retained belief is neither a current nor a recollected clear and distinct perception in my sense of these terms; it (merely) lingers in memory. Such beliefs originate in clear and distinct perception, but they are not based on clear and distinct perception in my stipulated sense. One cannot in general apply the truth rule to a lingering belief in a proposition, because one need not believe that the proposition was clearly and distinctly perceived.25 One can only apply the truth rule to a proposition that one takes to have been clearly and distinctly perceived. This is why Descartes claims that knowledge of the truth rule is sufficient for the unshakability specifically of propositions that one remembers having clearly and distinctly perceived.26 Since current clear and distinct perceptions are unshakable in any case, knowledge of the truth rule is sufficient for the unshakability of beliefs based on clear and distinct perception. In the context of the problem of the Cartesian circle, we may confine our attention to a more limited claim: that knowledge of the truth rule is a sufficient condition for the unshakability of beliefs based on clear and distinct perception, by the supposition that the skeptical hypothesis is true. This is what I have in mind when I write of “unshakability” later.

5. Knowledge of the Truth Rule as Securing Unshakability: A Conflict The passages from Meditation V, the Second Replies, and the letter to Regius do not explain how knowledge of the truth rule secures unshakability.

23. There is a related passage in Meditation III: “The only reason for my later judgement that they [simple propositions of mathematics] were open to doubt was that it occurred to me that perhaps some God could have given me a nature such that I was deceived even in matters which seemed most evident” (CSM 2.25: AT 7.35–36). Beliefs based on clear and distinct perception are doubtful only insofar as they are held by someone who does not have knowledge of the truth rule; such knowledge is sufficient to remove doubt. If doubt is not only able to dislodge belief but also necessary for belief to be dislodged, then this passage is equivalent to the claim that knowledge of the truth rule is sufficient for unshakability. 24. In the letter to Regius, Descartes claims that a demonstration of the existence of a nondeceiving God is sufficient to secure unshakability. This should be understood against the background of the passages from Meditation V and the Second Replies; such a demonstration is sufficient insofar as it enables one to demonstrate the truth rule. 25. One might believe on the basis of some inductive evidence that the proposition had been clearly and distinctly perceived. 26. Descartes is considering the unshakability of beliefs that one remembers having clearly and distinctly perceived, on the assumption that the memory is correct. Cf. Frankfurt 1962,

the cartesian circle

99

The material developed to this point permits a relevant inference. An unshakable belief has the psychological property that it cannot be dislodged by argument. If the skeptical supposition can dislodge recollected clear and distinct perceptions, and if knowledge of the truth rule results in the unshakability of recollected clear and distinct perceptions, then knowledge of the truth rule must be psychologically incompatible with the supposition that the skeptical hypothesis is true.27 Descartes writes in the Second Replies: Hence you see that once we have become aware that God exists it is necessary for us to imagine that he is a deceiver if we wish to cast doubt on what we clearly and distinctly perceive. And since it is impossible to imagine that he is a deceiver, whatever we clearly and distinctly perceive must be completely accepted as true or certain. (CSM 2.103: AT 7.144) Knowledge that an all-perfect God exists (and that deception is an imperfection) is psychologically incompatible with the supposition that God is a deceiver. Because Descartes claims that if God exists, clear and distinct perception could be defective only if God is a deceiver, he presumably holds that knowledge that an all-perfect God exists is psychologically incompatible with the supposition that clear and distinct perception is defective. The Second Replies therefore confirms the present interpretation.28 It remains to refine the thesis that “knowledge” of the truth rule is sufficient for unshakability. Current clear and distinct perceptions are psychologically irresistible. Belief in the truth rule is psychologically irresistible whenever it is a current clear and distinct perception. At times when the belief that whatever one clearly and distinctly perceives is true is psychologically irresistible, it is psychologically impossible to suppose that clear and distinct perception is defective. At least Descartes would take this to be psychologically impossible provided he assumes that the presence of an irresistible belief that p is psychologically incompatible with a concurrent supposition that ~p. Descartes relies on this assumption in the passage from the Second Replies quoted in the preceding paragraph. A current clear and distinct perception of the truth rule is psychologically incompatible with the supposition that the skeptical hypothesis is true, so that recollected clear

510–11, and 1970, 160–61, and Bennett, 1990, §§7, 12. Descartes does not directly confront the question of whether the unshakability of the memory belief that one did clearly and distinctly perceive the proposition can itself be secured. I believe there are difficulties here for any interpretation of Descartes, irrespective of whether “scientific knowledge” is conceived psychologically or epistemically. 27. Alternatively, knowledge of the truth rule, though compatible with the skeptical supposition, defeats the ability of the skeptical supposition to dislodge recollected clear and distinct perceptions. It will become apparent that Descartes does not envision this alternative. 28. Rubin has noted that this passage confirms the psychological interpretation (1977, 205).

100

reflection and the stability of belief

and distinct perceptions are unshakable at any time the truth rule is a current clear and distinct perception. (Because the irresistibility of current clear and distinct perception applies to both intuition and demonstration, this result holds for any time one intuits the truth rule or attends to its demonstration— for any time the truth rule is a current axiom or a current theorem.29 It is for this reason that Descartes does not maintain, either in paragraph fifteen of Meditation V or in the letter to Regius, that an intuition of the truth rule is required to secure unshakability. For expository purposes, I often assume that the truth rule is a theorem, not an axiom demonstrated, not intuited.) The explanation of why a current clear and distinct perception of the truth rule is psychologically incompatible with the skeptical supposition does not generalize to recollected clear and distinct perceptions. Because recollected clear and distinct perceptions are not psychologically irresistible, the recollection that one previously demonstrated the truth rule is not psychologically incompatible with the skeptical supposition. It is psychologically possible for the skeptical supposition to arise, and to dislodge recollected clear and distinct perceptions, at times when the truth rule is a recollected theorem.30 Recollected clear and distinct perceptions, unlike current clear and distinct perceptions, do not constitute “knowledge” of the truth rule for the purposes of securing unshakability. Recollected clear and distinct perceptions would be unshakable at all times for someone who always or continually attends to the demonstration of the truth rule.31 I believe Descartes would reject this technique for achieving unshakability at all times simply on the ground that it is not humanly possible to sustain such perpetual attention: “my nature is . . . such that I cannot fix my mental vision continually on the same thing, so as to keep perceiving it clearly” (CSM 2.48: AT 7.69; cf. CSM 2.43, 1.197: AT 7.62, 8A.9). Descartes claims in Meditation V that the recollection that one demonstrated the truth rule is sufficient for unshakability, even if one is no longer attending to the demonstration (cf. CSM 2.48: AT 7.70).32 There is a 29. In requiring that “the truth rule must be gone through in a single intellectual sweep, all held before the mind at once” (1990, §9), Bennett overlooks Descartes’ application of the doctrine of the irresistibility of clear and distinct perception to attending to a demonstration. It is the memory interpretation that exerts pressure in the direction of the requirement that a clear and distinct perception of the truth rule be compressed into a momentary intuition—a demonstration of the truth rule relies on memory of the earlier steps, and therefore apparently could not be invoked to validate memory. Cf. Stout 1929, esp. 463–67; Doney 1955, 328–29; and Frankfurt 1962, 508–9, and 1970, 158–59. 30. Cf. Feldman and Levison 1971, 496. Rubin overlooks this point (1977, 206–8). 31. Cf. Parsons 1972, 40, and Garns 1988, 89. Van Cleve 1979, n.31, and Bennett 1990, §9, suggest that this represents Descartes’ position. I do not see how this suggestion can accommodate the texts I proceed to cite. 32. For discussions of the bearing of this claim on the problem of the Cartesian Circle, see Doney 1970, esp. 393–96; Frankfurt 1970, 159–60, 172–77; Parsons 1972, 40–41, 43; Curley 1978, 104ff.; Gombay 1979, 495–97; Etchemendy 1981, 11–12, 18–19, 34–35; and Garns 1988, 88–89.

the cartesian circle

101

similar passage in the letter to Regius (cf. CSMK 147: AT 3.65).33 Descartes’ claim that a demonstration of the truth rule is sufficient for unshakability means that unshakability is secured if the truth rule is either a current theorem or (subject to a qualification to be discussed) a recollected theorem. As we have seen, however, the recollection that one demonstrated the truth rule does not secure unshakability.34 Our task is to reconcile this conflict.

6. A First Step in Resolving the Conflict: Two Senses of “Unshakability” The first step in the explanation is to locate a weakened notion of unshakability. I have characterized an unshakable belief as one that cannot be dislodged by argument. I now call this unshakability in the strong sense. A weaker sense of unshakability emerges in Meditation V. In paragraph fourteen, Descartes has explained that, if he lacks knowledge of the truth rule, he can entertain the skeptical supposition, a supposition that can undermine recollected clear and distinct perceptions. He writes in paragraph fifteen:

33. There is one apparent difference (apart from that discussed at note 24). Descartes claims in Meditation V that beliefs based on clear and distinct perception are unshakable, provided one recollects having clearly and distinctly perceived the truth rule; the proviso in the letter to Regius is that one recollect this “conclusion” (CSMK 147: AT 3.65)—not that one recollect having clearly and distinctly perceived the truth rule. Descartes writes in the Second Replies: “The question will now arise as to whether we possess the same firm and immutable conviction concerning these conclusions, where we simply recollect that they were previously deduced from quite evident principles (our ability to call them ‘conclusions’ presupposes such a recollection)” (CSM 2.104: AT 7.146). If we apply the parenthetical principle to the proviso in the letter to Regius, it is equivalent to the proviso in Meditation V. 34. Observing that “the skeptic . . . cannot doubt [the existence and benevolence of God] while he intuits the proofs of it,” B. Williams addresses the possibility that “the skeptic, ceasing to intuit the proofs, then reverts to objecting merely because he is no longer intuiting”: “the use of propositions one is not at that instant intuiting is a minimal structural condition on getting on at all in the acquisition of systematic knowledge, and . . . it would be unreasonable to spend all one’s time rehearsing the proofs of the general answer to skepticism” (1983, 349, and cf. 352n.13; cf. 1978, 206). Unshakability plays no essential role in B. Williams’ interpretation. It would be necessary to use previous clear and distinct perceptions for acquiring systematic knowledge whether unshakability is a goal of inquiry or not. Also, beliefs that are “cumulative” (1983, 345, and 1978, 202) and systematic could be dislodged by the skeptical supposition. Although B. Williams holds that scientific knowledge should be unshakable in the sense of “immune to being recalled into doubt” (1983, 349, and cf. 344, 345; cf. 1978, 202, 204), he does not explain how unshakability (as he has characterized it) can be achieved, or even how the “minimal structural condition” can be satisfied in the case of the truth rule. According to B. Williams, the skeptic is enjoined to continue to use or accept (cf. 1978, 200–206) the truth rule when it is a recollected clear and distinct perception. Belief in the truth rule, as distinct from irresistible belief in the truth rule, is compatible with the skeptical supposition, a supposition that can dislodge recollected clear and distinct perceptions, to include the truth rule itself. For a useful exposition and critique of B. Williams, see Stubbs 1980.

102

reflection and the stability of belief

Now, however, I have perceived that God exists, . . . and I have drawn the conclusion that everything which I clearly and distinctly perceive is of necessity true. Accordingly, even if I am no longer attending to the arguments which led me to judge that this is true, as long as I remember that I clearly and distinctly perceived it, there are no counterarguments which can be adduced to make me doubt it, but on the contrary I have true and certain knowledge of it. And I have knowledge not just of this matter, but of all matters which I remember ever having demonstrated, in geometry and so on. For what objections can now be raised? (CSM 2.48: AT 7.69–70) Subsequent to demonstrating the truth rule, “there are no counter-arguments which can be adduced to make [impellere] me doubt” either the truth rule itself, or other recollected theorems. The Latin ‘impellere’ can mean either ‘force’, ‘make’, ‘constrain’, or ‘compel’, on the one hand, or ‘cause’, ‘lead’, ‘bring’, or ‘induce’, on the other. These readings differ, though either yields a psychological account of unshakability. Doubt is a state that is able to dislodge belief. To claim that no counterarguments can be adduced to cause one to doubt recollected theorems is to claim that recollected theorems cannot be dislodged—that they are unshakable in the strong sense. Causing one to lose a belief differs from forcing one to lose a belief. Someone might cause a self-defense expert to relinquish some money, without forcing him to relinquish the money, if the self-defense expert does not avail himself of his means of preventing the loss. Similarly, an argument could cause one to lose a belief, without forcing one to lose the belief, if one possessed the means to prevent loss of the belief but failed to avail oneself of those means. The belief would nevertheless be unshakable, in the sense that one is able to prevent its being dislodged. I call this unshakability in the weak sense. Somewhat more precisely, a person’s belief is unshakable, in the weak sense, just in case the person possesses arguments that enable him to prevent the belief’s being dislodged by argument. A belief that is unshakable in the strong sense is unshakable in the weak sense, but not vice versa.35 To claim that no counterarguments can be adduced to force one to doubt recollected theorems is to claim that one can prevent recollected theorems from being dislodged— that they are unshakable in the weak sense. In the French edition of the Meditations, Descartes adds, following “For what objections can now be raised,” “to oblige me to call these matters into doubt” (CSM 2.48: AT 9A.56).36 Once one has demonstrated the truth rule, there are no arguments that can oblige (obliger) one to doubt recollected clear and distinct perceptions. Descartes’ point is that recollected clear and distinct

35. Of the commentators cited in note 11, only Gombay (1979, 492) is sensitive to this distinction. Bennett’s notion of “stability” is akin to unshakability in the strong sense: belief “that one won’t later be forced to give up” (1990, §1). 36. Haldane and Ross (1.184) do not translate the material from the French edition.

the cartesian circle

103

perceptions are unshakable because there are no objections that can force one to doubt them, and hence that one can prevent their being dislodged. This evidence that Descartes is concerned to achieve unshakability in the weak sense occurs in one of the two passages where Descartes claims that the recollection that one demonstrated the truth rule is sufficient for scientific knowledge.37 Apart from such textual details, the verb ‘impellere’ permits us to read Descartes as concerned with the weakened sense of unshakability. We should adopt that reading, if it enables us to find in Descartes an explanation of how recollecting that one demonstrated the truth rule secures unshakability.

7. A Second Step in Resolving the Conflict: Reproducibility The second step in the explanation is to observe that recollecting that one demonstrated the truth rule does secure unshakability in the weak sense, provided one retains the ability to reproduce the demonstration. Recollecting that one demonstrated the truth rule, though psychologically compatible with the supposition that the skeptical hypothesis is true, enables one to attain a state that is psychologically incompatible with that supposition, provided one retains the ability to reproduce the demonstration. The (attentive) exercise of the ability to reproduce the demonstration generates an irresistible belief in the truth rule, thereby preventing one from supposing that the skeptical hypothesis is true, and thus preventing one’s recollected clear and distinct perceptions from being dislodged. Someone who retains the ability to reproduce the demonstration of the truth rule possesses arguments—arguments that include the demonstration of the truth rule— that enable him to prevent recollected clear and distinct perceptions from being dislodged; these beliefs are unshakable in the weak sense. I am now in a position to add the required qualification to Descartes’ claim that recollecting that one demonstrated the truth rule is a sufficient condition for unshakability. It is sufficient for someone who retains the ability to reproduce the demonstration.38 I call the position sketched in this paragraph the 37. The second passage, in the letter to Regius, is susceptible to the same treatment: “There is conviction when there remains some reason which might lead [impellere] us to doubt, but scientific knowledge is conviction based on argument so strong that it can never be shaken by any stronger argument” (K 74: AT 3.65). Here too, we can understand ‘impellere’ in the sense of “to force.” 38. Garns speculates (1988, 97) that Descartes intends that whenever a meditator who has demonstrated the existence of a nondeceiving God and internalized the rules of Cartesian method considers the notion of God, or an omnipotent being, or the source of his being, he will automatically recall the proof that this being is a nondeceiver (98–99). (Garns does not embed this suggestion within the psychological interpretation. He endorses a nonpsychological solution to the problem of the circle—cf. 87–88.) Garns takes this position to be unsuccessful because it ignores the hypothesis that there exists an evil demon (97–99)—the notion of an evil demon would not automatically trigger the proof of the existence of a nondeceiving God (99).

104

reflection and the stability of belief

reproducibility account of how recollecting that one demonstrated the truth rule secures scientific knowledge. The reproducibility account commits Descartes to an asymmetry between two kinds of recollected clear and distinct perceptions. I call propositions that are essential to the demonstration of the truth rule basic propositions. The basic propositions include, among others, the principles about causation invoked in Meditation III in the course of the demonstration that God exists, and the propositions that God exists, that deception is an imperfection, and that God is no deceiver. We can think of any proposition as essential to its own demonstration, so that the truth rule is itself a basic proposition. Nonbasic propositions are not essential to the demonstration of the truth rule—a theorem in geometry would be an example. According to the reproducibility account, all recollected clear and distinct perceptions— whether they are beliefs in basic or nonbasic propositions—are rendered unshakable by the ability to reproduce the demonstration of the truth rule. The reproducibility account imposes no general requirement that one retain the ability to reproduce the demonstrations of propositions that one recollects having demonstrated. The ability to reproduce the demonstration of the truth rule, however, presupposes the ability to reproduce clear and distinct perceptions of each of the basic propositions—one cannot reproduce the demonstration of the truth rule without clearly and distinctly perceiving that God exists, that God is no deceiver, and so forth. Recollected clear and distinct perceptions of basic propositions constitute a special case; they cannot be rendered unshakable unless one retains the ability to reproduce their own demonstrations.39

This consideration is not compelling. The demon hypothesis places clear and distinct perception in doubt only if one supposes that the demon is “evil” in the sense of causing one to have a defective faculty of clear and distinct perception. Why not say, in the spirit of Garns’ speculation, that Descartes holds that this supposition automatically leads one to recall the proof that clear and distinct perception is not defective? A more serious difficulty is that Descartes nowhere maintains that there are circumstances in which the consideration of appropriately related propositions or notions automatically triggers reproduction of a given demonstration. Laporte, citing the penultimate paragraph of Meditation IV, holds a weaker version of the position: by “attentive and repeated meditation” (CSM 2.43: AT 7.62), a Cartesian meditator can habituate himself to remember the conclusion that God is no deceiver in order to check the skeptical supposition (1950, 161). If the truth rule is merely a recollected theorem, however, it will not be psychologically irresistible and will not suffice to block the skeptical supposition. 39. Etchemendy considers a strengthened version of the reproducibility account, on which our knowledge that we have the ability to reproduce the demonstration of the truth rule (rather than our merely having the ability) protects us from metaphysical doubt. Etchemendy then rejects this position, appealing to an argument designed to show that such knowledge would be otiose (cf. 1981, 19). An adaptation of his argument suggests that the ability to reproduce the demonstration of the truth rule would itself be otiose: if the ability to reproduce the demonstration of the truth rule secures unshakability, then the ability to reproduce the demonstration of any proposition that we recollect having clearly and distinctly perceived would equally secure that proposition’s unshakability. This overlooks the point that Descartes seeks to secure,

the cartesian circle

105

The reproducibility account goes beyond anything Descartes directly says insofar as it requires that achieving unshakability in the weak sense depends on retaining the ability to reproduce the demonstration of the truth rule. We need to inquire whether it is plausible to attribute this view to Descartes. There is no obstacle in principle to a person’s retaining the ability to reproduce a particular demonstration. It is no objection that particular persons might lack this ability. Any account of Descartes’ attempt to remove the doubt about clear and distinct perception will assign a role to the demonstration of the truth rule. It is no objection that some persons are not capable of comprehending this demonstration. Descartes is trying to show how knowledge is possible for humans with ordinary cognitive endowments (cf. CSM 1.111–12: AT 6.1–3), not that any human—even one with subnormal cognitive abilities—can achieve knowledge. The present version of the psychological interpretation takes “knowledge,” that is, scientific knowledge, to require unshakable belief. Descartes’ claim is that unshakable belief is possible for someone who does retain the ability to reproduce the demonstration of the truth rule. If some persons do not retain this ability, this at most shows that such persons do not have unshakable beliefs.

8. Passages That Bear Directly on the Reproducibility Account I turn to textual obstacles to the reproducibility account. Of the many passages in which Descartes claims that unshakability depends on “knowledge” of the truth rule, only two mention that the recollection that one demonstrated the truth rule secures unshakability. The issue is whether recollecting that one demonstrated the truth rule constitutes knowledge of the truth rule (for the purpose of securing unshakability) even if one is no longer able to reproduce the demonstration. In the Meditation V passage, Descartes claims that various beliefs are unshakable “even if I am no longer attending to the arguments which led me to judge that this is true, so long as I remember that I clearly and distinctly perceived it.” The letter to Regius contains similar language. We can distinguish two situations in which one remembers having demonstrated a proposition, without at the same time attending to the demonstration. In situation (i), one is not attending to the demonstration because one has forgotten it; one has not retained the ability

insofar as humanly possible, the (concurrent) unshakability of all recollected clear and distinct perceptions. In the absence of the ability to reproduce the demonstration of the truth rule, such unshakability can be achieved only if one has the ability (i) to reproduce the demonstrations of every recollected clear and distinct perception and (ii) to do so concurrently. The human mind’s finite capacity—as Etchemendy should grant (cf. 8)—is incompatible with (ii). Also, for reasons just explained in the text, (i) is a much more burdensome condition than that required by the reproducibility account. (See also §9.)

106

reflection and the stability of belief

to reproduce the demonstration. In situation (ii), one is not attending to the demonstration even though one has not forgotten it; one remembers the demonstration in the sense that one retains the ability to reproduce it, but is not exercising that ability at the time of the recollection. It would be an obstacle to the reproducibility account if Descartes held that recollecting, in situation (i), that one demonstrated the truth rule secures unshakability. Descartes’ language, however, is compatible with (ii). In the two passages under consideration, Descartes states the conditions for securing the unshakability specifically of basic propositions and generalizes the result to nonbasic propositions. Descartes writes in paragraph fifteen of Meditation V: “I have drawn the conclusion that everything which I clearly and distinctly perceive is of necessity true. Accordingly, even if I am no longer attending to the arguments which led me to judge that this is true, as long as I remember that I clearly and distinctly perceived it, there are no counter-arguments which can be adduced to make me doubt it.” Once one has demonstrated the truth rule, belief in the truth rule, a basic proposition, is unshakable. Descartes extends this result: “And I have knowledge not just of this matter [the truth rule], but of all matters which I remember ever having demonstrated, in geometry and so on.” Nonbasic propositions are also unshakable, once one has demonstrated the truth rule. In the letter to Regius, Descartes also distinguishes between basic and nonbasic propositions, and holds that the unshakability of both is secured by remembering that one demonstrated basic propositions, even if one no longer attends to their demonstration (cf. CSMK 147: AT 3.65). In the two passages in which Descartes directly addresses the question of what qualifies as “knowledge” of the truth rule for the purpose of securing unshakability, Descartes applies language compatible with situation (ii) specifically to basic propositions. There are seven passages in which Descartes claims that the unshakability of recollected clear and distinct perceptions depends on knowledge of the truth rule, without distinguishing between the unshakability of basic and nonbasic propositions. Six of these passages contain the language, or a modest variant of the language, of remembering having clearly and distinctly perceived a proposition without at the same time attending to the clear and distinct perception (cf. CSMK 353; CSM 1.197, 2.48, 100, 171, 309: AT 5.178, 8A.9–10, 7.69–70, 140, 245–46, 460). These six passages are compatible with situation (ii). Some of these passages use supplementary language strongly suggestive of situation (ii) rather than situation (i). Descartes writes: “as soon as I turn my mind’s eye away from the proof, then in spite of still remembering that I perceived it very clearly, I can easily fall into doubt about its truth, if I am without knowledge of God” (CSM 2.48: AT 7.70). Only one passage is prima facie incompatible with the reproducibility account. Descartes writes in the Second Replies: There are other truths which are perceived very clearly by our intellect so long as we attend to the arguments on which our knowledge of

the cartesian circle

107

them depends; and we are therefore incapable of doubting them during this time. But we may forget the arguments in question and later remember simply the conclusions which were deduced from them. The question will now arise as to whether we possess the same firm and immutable conviction concerning these conclusions, when we simply recollect that they were previously deduced from quite evident principles. . . . My reply is that the required certainty is indeed possessed by those whose knowledge of God enables them to understand that the intellectual faculty which he gave them cannot but tend towards the truth. . . . This point was explained so clearly at the end of the Fifth Meditation that it does not seem necessary to add anything further here. (CSM 2.104–5: AT 7.146) The language of remembering a conclusion though we have forgotten the arguments for it suggests that Descartes is envisioning situation (i). I do not think this is strong evidence against the reproducibility account. First, if the reproducibility account is correct, Descartes’ position has the complexity of involving the asymmetry noted previously (§7)—recollected clear and distinct perceptions of nonbasic propositions, unlike basic propositions, can be rendered unshakable even if one does not retain the ability to reproduce their demonstrations. The Second Replies passage is not incompatible with this interpretation, if Descartes is focusing on nonbasic propositions. Second, the final sentence of the passage refers the reader to the close of Meditation V for a fuller statement of Descartes’ position. We have seen that in Meditation V Descartes’ discussion of the unshakability specifically of basic propositions is compatible with situation (ii). Finally, the Second Replies passage is an exception insofar as it seems to envision (i). The eight other discussions of securing unshakability—the six cited in the preceding paragraph, together with the two passages in which Descartes claims that the recollection that one demonstrated the truth rule is sufficient for unshakability—are all compatible with (ii).40 Indeed, in light of these eight passages,

40. Haldane and Ross translate a portion of Principles I.13: “But since it cannot always devote this attention to them [when it remembers the conclusion and yet cannot recollect the order of its deduction], and conceives that it may have been created of such a nature that it has been deceived even in what is most evident, it sees clearly that it has great cause to doubt the truth of such conclusions” (HR 1.224, emphasis added). They translate a portion of Meditation V: “As I often recollect having formed a past judgment without at the same time properly recollecting the reasons that led me to make it, it may happen meanwhile that other reasons present themselves to me, which would easily cause me to change my opinion” (1.183, emphasis added). The language of remembering a conclusion without recollecting its demonstration suggests that the demonstration has been forgotten, that Descartes is envisioning situation (i). In the first passage, the bracketed material renders the French: “sans prendre garde à l’ordre dont elle peut etre démontrée” (AT 9B.30–31). In the second passage, the Latin verb is ‘attendere’ (AT 7.69). Neither text carries any suggestion that the demonstration has been forgotten. Tlumak is mistaken in his claim that “Descartes repeatedly insists that, once the existence of a good God . . . is acknowledged, we are certain of the conclusion of a proof we correctly remember having

108

reflection and the stability of belief

it seems permissible to understand the Latin ‘oblivisci’ (AT 7.146) and the French ‘oublier’ (AT 9A.115) less literally as “lose sight of” or “be unmindful of.” This reading of the Second Replies is compatible with (ii). The overall textual evidence strongly suggests that Descartes did not hold that recollected clear and distinct perceptions are unshakable once one has demonstrated the truth rule, even if one is no longer able to reproduce the demonstration.41

9. Immutability and Permanence There remain textual obstacles to the reproducibility account from another quarter. Descartes’ notion of unshakability—or firmness or solidity—is one of a cluster of interconnected concepts. Descartes writes of beliefs that are both firm and immutable or unchangeable (CSM 2.103, 104: AT 7.145, 146) and of beliefs that are immutable (CSM 2.289: AT 7.428)—in contrast to beliefs that are mutable (CSM 2.48: AT 7.69). He also writes of beliefs that are both firm and lasting (or permanent) (CSM 2.12: AT 7.17)—in contrast to beliefs that are fluctuating (CSM 1.14: AT 10.368). It seems clear that immutable belief and permanent belief are themselves objectives of inquiry. Permanence is prima facie distinct from immutability, understood literally. Whereas a belief that is literally immutable must be permanent, a belief could be permanent without being immutable. A belief, once one holds it, could be mutable, even though it does not in fact change. Although Descartes places more weight on immutability, permanence is prominent in the first paragraph of the Meditations. The reproducibility account owes us an explanation of the interconnections between unshakability, immutability, and permanence. In particular, it needs to be shown that securing unshakability is compatible with securing immutability and permanence. I begin with immutability. (In what follows, I construe ‘immutable’ to mean “immutable specifically by argument.”) A belief that is unshakable in the strong sense that it cannot be dislodged could be immutable.42 A belief

clearly and distinctly perceived, even though we cannot reproduce its premises” (1978, 49). Descartes says this at most once, in the Second Replies. 41. The reproducibility account can be generalized: if the ability to reproduce the demonstration of the truth rule is sufficient for unshakability, the ability to produce a demonstration of the truth rule ought to be sufficient for unshakability. Descartes nowhere claims that a (current or previous) demonstration of the truth rule is necessary for unshakability. He does say that “firm and immutable conviction concerning” recollected clear and distinct perceptions “is indeed possessed by those whose knowledge of God enables them to understand that the intellectual faculty which he gave them cannot but tend towards the truth” (CSM 2.104–5: AT 7.146). Unshakability requires the ability to demonstrate the truth rule, not that the truth rule has been demonstrated. 42. “Could”—rather than would—for reasons that emerge later in §9: immutability, unlike unshakability, is a continuing property of a belief.

the cartesian circle

109

that is unshakable in the weak sense can be dislodged, if one does not exercise the ability to prevent its being dislodged. A belief that can be dislodged, even if it is not dislodged, is not literally immutable. Unshakability in the strong sense is compatible with literal immutability but cannot itself be achieved; unshakability in the weak sense can be achieved but is not compatible with literal immutability. The only escape from this dilemma is to locate some nonliteral, technical sense of the term ‘immutable’ that is compatible with mere unshakability in the weak sense. Although there are a number of passages that provide discursive treatments of “unshakability,” “immutability” (as it applies to human belief or knowledge) does not receive equal treatment. The closest approximation to a gloss occurs in the Second Replies: For the supposition which we are making here is of a conviction so firm that it is quite incapable of being destroyed; and such a conviction is clearly the same as the most perfect certainty. But it may be doubted whether any such certainty, or firm and immutable conviction, is in fact to be had. (CSM 2.103: AT 7.145) Descartes identifies firm conviction with perfect certainty, and perfect certainty with conviction that is firm or unshakable and immutable. What is the force of this additional condition in the account of scientific knowledge?43 Suppose that, at time t, one believes that p, remembers clearly and distinctly perceiving that p, and retains the ability to reproduce a demonstration of the truth rule; suppose, in other words, that the belief that p is unshakable at t. The unshakability of a belief at a time t does not guarantee its unshakability at a subsequent time t′. There are various possibilities: (a) one might have forgotten the belief by t′; (b) one might retain the belief at t′, but have forgotten that it was clearly and distinctly perceived; or (c) one might, at t′, retain the belief and remember that it was clearly and distinctly perceived but have forgotten the demonstration of the truth rule. In case (a), one no longer has a belief, much less an unshakable belief, in the proposition at t′. In case (b), one could not apply the truth rule to the belief at t′. In case (c), one does not, at t′, satisfy a necessary condition for the unshakability of any recollected clear and distinct perception. Unshakable beliefs can withstand being dislodged by argument. In cases (a–c), belief is not dislodged by argument; rather, unshakable belief is lost due to forgetfulness. Possession of an unshakable belief that p at t does not guarantee continued possession of an unshakable belief that p. (This conclusion applies to unshakability both in the strong and weak senses.) Immutability, by contrast,

43. Gombay—the only commentator I know who attempts to distinguish firmness and immutability—identifies immutability, rather than firmness, with unshakability (cf. 1979, 492–94, 498–500). I do not see how this interpretation can accommodate the evidence from the Second Replies and the Search cited in my earlier discussion of the firmness metaphor.

110

reflection and the stability of belief

carries the connotation of being a continuing property of a belief. Because the unshakability of a belief is relative to a specific time, unshakability does not entail immutability. I suggest that the immutability requirement represents Descartes’ recognition of a gap between unshakability at a time and continuing unshakability. We can think of the requirement as imposing such further conditions as are necessary for the possession of an unshakable belief in a proposition to guarantee the continued possession of an unshakable belief in that proposition. This requirement adds to the concept of scientific knowledge, without requiring that a belief be literally immutable. We can explain why Descartes does not emphasize the distinction between unshakability and immutability. In each of cases (a–c), unshakability is insufficient for continuing unshakability due to the possibility of forgetfulness.44 Forgetfulness is a matter of individual variation in cognitive ability or performance; it is not intrinsic to human nature. Individual limitations in forgetfulness do not preclude the possibility of human knowledge, even if such knowledge requires continuing unshakability (in the weak sense).45 I turn to permanence in belief. Even continuing unshakability in the weak sense is compatible with impermanence; a belief with continuing unshakability in the weak sense can be lost if one fails to avail oneself of one’s means of preventing the belief’s being dislodged. The issue is whether continuing unshakability in the weak sense is also compatible with permanence. Insofar as recollected clear and distinct perceptions are dislodged by the skeptical supposition, permanence requires that one never suppose that the skeptical hypothesis is true. It is tempting to think that securing the permanence of recollected clear and distinct perceptions—much as securing their unshakability in the strong sense—therefore requires continual

44. Distinctions between basic and nonbasic propositions, between various objectives of inquiry, and between various kinds of forgetfulness are necessary in order to evaluate whether nonforgetfulness is, for Descartes, itself an objective of inquiry or a requirement for scientific knowledge. There is an additional case: (d) one might have forgotten the demonstration of p by t′. Since this reduces to case (c) if p is a basic proposition, I restrict (d) to propositions that are not basic. In case (d), the memory loss does not preclude continuing unshakability through time t′. In cases (a–c), the memory loss does preclude continuing unshakability through time t′. In cases (a–d), the memory loss is compatible with the unshakability of the belief that p at the earlier time t. As far as I can see, only in case (a) is the memory loss automatically incompatible with the permanence of the belief that p. For previous discussions of forgetfulness, see Feldman and Levison 1971, 496; Kenny 1971, 498; Tlumak 1978, 49; and Markie 1986, 65–69. 45. Perhaps some forgetfulness of the sort in cases (a) or (b) is intrinsic to human nature. This does not show that continued unshakability for the greater part of those beliefs that were once clearly and distinctly perceived is an unrealistic goal; it is only unrealistic for individuals who have systematic failures of memory, a condition that is not intrinsic to human nature. Cases (a) and (b) are incompatible with the continued unshakability specifically of the proposition p in question. It is case (c) alone that is incompatible with the continued unshakability of any belief, but it is humanly possible to remember the demonstration of the truth rule. This point undermines an objection that Feldman and Levison (1971, 496, third paragraph) direct at Kenny.

the cartesian circle

111

attention to the demonstration of the truth rule. This overlooks Descartes’ claim that it is not humanly possible continually to attend to any one matter so as to perceive it clearly. Just as one’s own nature prevents one from continually attending to the demonstration of the truth rule, one’s own nature prevents one from continually attending to the skeptical hypothesis. The ability to reproduce the demonstration of the truth rule need not be continuously exercised, even to achieve permanence in belief, because the supposition that the skeptical hypothesis is true will not itself continually recur.46 What about those occasions when the skeptical supposition does recur? Some commentators have observed that the supposition can then be dislodged by exercising one’s ability to reproduce the demonstration of the truth rule.47 This is a technique for restoring beliefs dislodged by the skeptical supposition by dislodging the skeptical supposition itself. Restoring a dislodged belief, however, does not achieve permanence in belief; it minimizes the loss or the impermanence—much as a self-defense expert might exercise his ability to recover his money after relinquishing it. Permanence in belief requires the preventative or preemptive exercise of the ability to reproduce the demonstration of the truth rule. A self-defense expert’s ability to prevent relinquishing his money depends on his ability both to recognize that the loss of his money is impending and to react defensively, on a timely basis. Similarly, the preemptive exercise of the ability to reproduce the demonstration of the truth rule depends on the ability both to recognize that the supposition that the skeptical hypothesis is true is impending and to reproduce the demonstration of the truth rule, on a timely basis. Impermanence due to the skeptical supposition can be avoided by such preemptive exercise of one’s ability to reproduce the demonstration of truth rule. Although this does not show that permanence can be achieved, it does show that there is nothing intrinsic to the reproducibility account to preclude achieving permanence. In sum, I see no significant textual obstacle to an interpretation on which Descartes holds that achieving unshakability depends on retaining the ability to reproduce the demonstration of the truth rule. This completes my exposition of the textual basis for a version of the psychological interpretation that incorporates the reproducibility account. Before closing, I offer some brief remarks concerning philosophical objections that might be raised to Descartes’ position as characterized by the psychological interpretation.

46. I assume that a mere disposition to suppose that the skeptical hypothesis is true would not be sufficient to dislodge occurrent recollected clear and distinct perceptions; in order to do so, the supposition would have to be occurrent. Cf. Descartes’ language at AT 7.25: CSM 2.25. 47. Cf. Laporte 1950, 161; Larmore 1984, 71; Cottingham 1986, 72; and Garns 1988, 98–99.

112

reflection and the stability of belief

10. Philosophical Objections to the Psychological Interpretation The demonstration of the truth rule begs the question against the skeptical supposition. That it does beg the question will be apparent, even to persons who have demonstrated the truth rule and recollect having done so, if they review their prior argumentative procedure. Such a person will also know that a current clear and distinct perception of the truth rule has the result that one irresistibly believes the truth rule, even though it is the conclusion of a question-begging argument. To proceed, under these conditions, to reproduce the demonstration of the truth rule seems akin to knowingly taking a pill, or knowingly submitting to a hypnotic spell, that induces an irresistible belief for which one lacks good evidence.48 This is knowingly to enter an epistemological illusion. The present version of the psychological interpretation does not succumb to this objection. Descartes is trying to show how scientific knowledge can be achieved. Although one cannot avoid believing the truth rule when one does reproduce its demonstration, one can avoid reproducing its demonstration. Someone who has the continuing ability to reproduce the demonstration of the truth rule can decline to exercise that ability. Such a person has nevertheless achieved continuing unshakability in the weak sense, and hence scientific knowledge, any scruples about the illusion notwithstanding. The objection can be put in a slightly different form. Would not someone who, under the conditions outlined, proceeded to reproduce the demonstration of the truth rule in order to restore a dislodged belief, or to preempt impermanence, be a party to the illusion? Or consider someone who supposes that the skeptical hypothesis is true. It comes to his attention that there is an argument, such that if he attends to the argument, he will irresistibly believe that the skeptical hypothesis is false. He wonders whether it is not in the nature of the case that the argument, whatever its details, will beg the question against the skeptical supposition. He satisfies himself that, inevitably, it will beg the question. This person’s beliefs are not yet unshakable. Would he not be a party to the illusion if he proceeded to attend to the argument for the truth rule for the first time? I speculate that Descartes would respond with reference to the costs of not entering the illusion. The illusion consists in the irresistible belief in the truth of a proposition for which one lacks good evidence. Any clear and distinct perception is implicated in the illusion. There is no clear and distinct perception for which one has good evidence, if good evidence requires a non-question-begging argument against the skeptical supposition. Avoiding the illusion requires that one decline to exercise one’s faculty of clear and

48. Cf. Rubin 1977, 208; B. Williams 1978, 207; Gombay 1979, 495; and Bennett 1990, §13.

the cartesian circle

113

distinct perception altogether. Descartes holds that whereas clear and distinct perception is internally coherent, sense-perception is internally incoherent— sense-perception on its own generates conflicting beliefs. Clear and distinct perception resolves these conflicts by sustaining one of the conflicting beliefs and correcting the other. The resolution is effected in virtue of an asymmetry in the psychological properties of the faculties: whereas clear and distinct perception is psychologically irresistible, sense-perception generates suppressible inclinations to hold beliefs.49 In declining to exercise the faculty of clear and distinct perception, one deprives oneself of the means for resolving the conflicts that arise within sense-perception. The resulting doxastic system would be inherently, and ineliminably, unstable.50 This instability could be avoided only by declining to exercise the faculty of sense-perception as well as that of clear and distinct perception. The cost of entering the illusion must be weighed against the cost of not doing so—ineliminable instability, or declining to use one’s cognitive faculties. It might be felt that these observations are not responsive to the underlying point of the objection—that the psychological response to the problem of the circle is of mere psychological, and no epistemic, significance.51 This objection is likely to be offered by those who think that an adequate solution to the problem of the circle must be epistemic in character. A complete reply would require a detailed comparison of the strengths and weaknesses of the epistemic and psychological interpretations. In my view, all known versions of the epistemic interpretation either fail to acquit Descartes of begging the question, acquit him of that charge only by misconstruing the question at issue, or collapse (under scrutiny) into versions of the psychological interpretation. Space does not permit me to defend this view. I can nevertheless state a partial reply: Descartes’ interest is in scientific knowledge as he conceives it; he does offer purely psychological characterizations of this and related notions that provide the materials for the psychological response to the problem of the circle.52 The texts provide an alternative to the epistemic account. I do take note of a strategy for accommodating passages that seem to require the epistemic interpretation. Much as one can irresistibly believe that whatever one clearly and distinctly perceives is true, one can irresistibly believe that the truth rule provides a good (or even conclusive) reason not to doubt beliefs based on clear and distinct perception. One can irresistibly believe this if one also clearly and distinctly perceives, and hence irresistibly believes, relevant epistemic principles about the relationship

49. I defend the attribution to Descartes of these doctrines in my 1990 [this volume ch. 2], §§2–3. 50. It would also lack comprehensiveness (cf. note 22). 51. Cf. Gewirth 1941, 379; Frankfurt 1965, 153, and 1970, 170–71; M. Wilson 1978, 133; and Markie 1986, 44. 52. I do not see how Markie can claim that there is “no textual support” for “purely psychological accounts of how reasons for doubt are ruled out” (1986, 44).

114

reflection and the stability of belief

between good reasons for a belief and the likelihood that the belief is true. More generally, from the perspective of the psychological interpretation, we can think of passages suggestive of the epistemic interpretation as implicitly embedded within the propositional attitude “I irresistibly believe that . . .” or as reports of what can be irresistibly believed.53 The details require development.

11. The Significance of the Psychological Interpretation The psychological interpretation and the replies I have sketched to the philosophical objections to the position it attributes to Descartes rely on Cartesian doctrines about the psychological properties of the cognitive faculties. Some might conclude that the psychological interpretation should be dismissed, on the ground that it makes Descartes’ position depend on an accidental or contingent fact of human psychology—that current clear and distinct perception is psychologically irresistible. This might be a reason for rejecting the psychological response to the problem of the circle. It is not a reason for rejecting the psychological interpretation, the attribution of the psychological response to Descartes. To the contrary, the considerable textual merit of this interpretation suggests that Descartes’ rationalism cannot be understood apart from the doctrine that clear and distinct perception, unlike sense-perception, is psychologically irresistible.54 (The interesting historical question is why Descartes insisted on this psychological doctrine.55) Indeed, I argue elsewhere that, for Descartes, the superiority or priority of reason or clear and distinct perception to sense-perception ultimately rests on the greater irresistibility of reason.56 I am inclined to think that the psychological properties of the cognitive faculties play a more essential role in seventeenth- and eighteenth-century epistemology than is generally understood.57 I believe that both Descartes and Hume adopt doxastic objectives characterized in psychological terms— objectives that relate to such properties as the permanence, unshakability, and stability of belief. In this sense, Descartes and Hume are engaged in a

53. Cf. Rubin 1977, 197–98, and my 1990, §6. Some passages that, in translation, seem to require an epistemic interpretation are susceptible to alternative translations that are more hospitable to the psychological interpretation. Such passages can be accommodated without recourse to the strategy I have outlined; on this point, cf. note 10. 54. Cf. Larmore 1984, 61–62. 55. For some attempted explanations, see Kenny 1968, 185; Doney 1970, 399–400; Frankfurt 1970, 164; and my 1990, §4. An adequate explanation would play a role in a more complete assessment of the merit of the psychological interpretation of Descartes’ position on the problem of the circle. 56. Cf. my 1990, esp. §§1–6. 57. For a more developed version of the themes in this paragraph, cf. my 1990, §7.

the cartesian circle

115

common epistemological project. An account of how to achieve such psychologically defined objectives is inextricably linked to a conception of the psychological properties of the cognitive faculties. It is in their conceptions of these properties that Descartes and Hume diverge: for Descartes, only reason generates psychologically irresistible beliefs; for Hume, irresistible beliefs result from sense-perception, memory, and causal inference, as well as from reason, or intuition and demonstration.58 This contrast is crucial to the difference between Descartes’ rationalism and Hume’s empiricism.59

58. Cf. Treatise 8, 31, 153, 225. 59. I thank Jonathan Bennett for access to a draft of his 1990 prior to its publication. I am grateful to Paul Boghossian, Richard Brandt, Stephen Darwall, Allan Gibbard, Gideon Rosen, Lawrence Sklar, William Taschek, David Velleman, Nicholas White, and Stephen Yablo for helpful discussion. I am especially grateful to John Cottingham for detailed written comments and suggestions.

116

reflection and the stability of belief

4 Sextus, Descartes, Hume, and Peirce On Securing Settled Doxastic States

1. Introduction sextus empiricus and Peirce are famously philosophers for whom the objective of inquiry or investigation is characterized in psychological terms—for Sextus, as quietude or tranquility, and for Peirce, as the settlement or fixation of belief. I maintain that Descartes and Hume also characterize the goal of inquiry psychologically, in terms of such notions as unshakability and equilibrium in belief, and that their approaches to assessing belief-forming methods with reference to psychological notions have affinities with those in Sextus and Peirce. Methods of forming doxastic states are assessed by each of the four figures with reference to effectiveness in achieving a settled condition in those states. My argument depends on identifying important lines of thought in Descartes and Hume, tendencies in their thinking that are present and not dismissed, that place a premium on settled doxastic states. In this chapter, I seek to bring to light these lines of thought. In the interest of presenting an overview of Descartes and Hume, against the backdrop of Sextus and Peirce, I provide a partial defense of admittedly contentious interpretive claims, ones I attempt to defend more fully elsewhere. I locate in Hume the claim that one ought to seek doxastic states that are settled. I locate in Descartes the claim that one ought to seek beliefs that are not only settled, but also incapable of being shaken or unsettled. Achieving a settled condition is necessary, but not sufficient, for achieving a settled condition that cannot be unsettled. The Cartesian objective is more ambitious, and the Humean more modest. I maintain that the value placed on settled doxastic states has a naturalistic foundation in three of the figures;

117

we find in Peirce, Hume, and Sextus, though not in Descartes, the claim that unsettled states are unpleasant, and hence to be avoided. It is Hume, however, who is odd man out in a different respect; unlike the other figures, he claims to be pessimistic that the objective of inquiry he identifies can be achieved. To make interpretive headway, we need to consider the notion that the objective of inquiry might be characterized in psychological terms. Belief exhibits two kinds of characteristics.1 In the first place, belief is typically associated with a first-order disposition, or set of dispositions, to behave in particular ways in particular conditions. This first characteristic focuses on the effects or outputs of belief. These include other internal states, as well as external behavior. In the second place, belief is typically associated with a second-order disposition to regulate one’s belief that a proposition is true, and hence the first-order dispositions associated with belief, by (what one takes to be) evidence or indicators of truth.2 This second characteristic focuses on the causes or inputs of belief. There is room for controversy about the relationship between these characteristics. At one extreme, it might be held that possession of the first-order disposition is sufficient for belief, so that one could believe a proposition even in the absence of the second-order disposition to regulate one’s first-order dispositions by evidence of the truth. I call this the output model of belief. At an opposing extreme, it might be held that the second-order disposition is constitutive of belief, so that the presence of the first-order disposition is not sufficient for belief, unless the second-order disposition is also present.3 I call this the input model of belief. An example might help to make vivid the differences between these models. Suppose someone possesses the first-order dispositions associated with the belief that p. We present strong evidence that the belief has been acquired and sustained by a highly unreliable method. Though the person appreciates this evidence, he is not moved or bothered by it, and indeed continues to possess the first-order dispositions. From the perspective of the output model, the person, in possessing the first-order dispositions, maintains the belief that p. From the perspective of the input model, the person’s attitude toward p is not belief, but rather something else. Let us call it blind faith that p; the person regards p as true, without possessing the second-order disposition to regulate his first-order dispositions by evidence of their truth. I do not intend to adjudicate the issues raised by the competing extremes. I simply take it that any account of the matter will both grant that belief is typically associated with

1. For a similar formulation, see Leon 1992, esp. 299–302. 2. This formulation is adapted from Velleman 1992, 14, including note 24. Also cf. Leon 1992, 300. I owe the distinction that follows to Velleman, in conversation. 3. This position is found in Velleman 1992, 10–15, and 1996, 707–11. Also cf. Humberstone 1992, 73–75, and Kobes 1992. Though Leon begins by taking the two features as “central and connected characteristics” of belief (1992, 299), his position at 304 seems to approach that of Velleman.

118

reflection and the stability of belief

appropriate effects and causes and also seek to explain the interrelations among these characteristics of belief. Let us return to the second-order disposition to regulate one’s belief that a proposition is true by evidence of its truth. The characterization of this disposition provides a sense in which belief “aims” at the truth; when we believe a proposition, we accept it as true with the aim that whether or not we believe it should be responsive to its truth, with the aim that belief should be sensitive to what is really true.4 And if belief aims at the truth, this might seem not to leave room for the objective of inquiry to be something else, something other than truth, such as tranquility, settlement, equilibrium, or unshakability. Short of compelling evidence, we ought not suppose that a historical figure would deny that, in having belief, we seek to regulate what we regard as true by evidence of truth. Even if such an aim is not constitutive of belief, it seems closely associated with it. Consider the example of the person who has blind faith that p. At the least, we are inclined to look for some special psychological explanation of his indifferent response to evidence. Perhaps, for example, the person is so invested in regarding p as true as to fall into self-deception about the evidential state of affairs.5 So it remains to understand how the objective of inquiry can be something other than truth. A possible answer is that truth and a favored objective characterized in psychological terms—let’s choose “stability”—are coordinate aims that trade off against each other. They might be related as objectives in something like the way that such features as simplicity, fruitfulness, and scope are sometimes thought to trade off as criteria of theory-acceptance. It is not clear, however, that we can reasonably regard “belief aims at the truth” and “belief aims at stability” as coordinate claims. Even if aiming at the truth is not constitutive of belief, aiming at the truth seems to be much more tightly related to our concept of belief than is aiming at stability. A better answer is that we can possess yet higher-order dispositions than aiming at the truth. An analogy to other activities might be helpful.6 Playing a game, such as chess, is typically associated with a set of first-order dispositions, to move pieces in accordance with the rules, and also a second-order disposition, to move pieces in accordance with the rules with the aim of winning the game. One’s engagement in the activity of chess, however, might itself be directed by a higher-order desire, for example, to have fun, or to develop skill, or to show off—by playing chess. For the sake of achieving these other aims, one is aiming to win. So we can think of activities—such as moving chess pieces, or regulating belief—as being directed by a hierarchy of dispositions. Of course, in playing chess and in regulating belief, there

4. This formulation is based on Velleman 1992, 13–15, and 1996, 709. Also see Leon 1992, 299, 300. 5. I owe this point to David Velleman. 6. Analogies to chess are used by Velleman 1996, 713–14.

sextus, descartes, hume, and peirce

119

need not be a higher-order objective than winning and aiming at the truth, respectively, but there might be. We might, for example, possess a third-order desire to possess the firstand second-order dispositions characteristic of belief for the sake of achieving stability in doxastic states (or, at least, we might possess a third-order desire to achieve stability in doxastic states in the course of aiming at the truth). Sextus represents the skeptic’s initial position in precisely this way: “the . . . Skeptic . . . set out to philosophize with the object of passing judgement on the sense-impressions and ascertaining which of them are true and which false, so as to attain quietude thereby” (PH I.26).7 Sextus portrays the skeptic as setting out to achieve true belief for the sake of, or as a means to, the higher-order objective of quietude. The possibility of higher-order desires explains how inquiry might have two distinct objectives. Viewing “aiming at truth” and “aiming at stability” as dispositions of different orders is also illuminating in application to Peirce. How could Peirce hold that the objective of inquiry is something other than truth? The answer is that Peirce takes the desire for settled belief as a higher-order disposition which directs the lower-order disposition to regulate one’s belief that a proposition is true by evidence of its truth. In Peirce’s view, reflection on the consideration that it is accidental that the employment of the method of authority and the a priori method have led us to form particular beliefs, and not others, will unsettle those very beliefs (cf. CP 5.381–83). It will unsettle those beliefs precisely because such reflection finds that they have been “determined by [a] circumstance extraneous to the facts” (5.383). This argument presupposes that belief characteristically aims at the truth. Were this not the case, the line of reflection Peirce describes would not be unsettling. Peirce endorses the presupposition: “A man . . . wishes his opinions to coincide with the fact, and . . . there is no reason why the results of those three first methods should do so” (5.387). Here, Peirce begins with a formulation of the idea that belief aims at the truth, and then assesses the methods of tenacity and authority, and the a priori method, with reference to it. In Peirce’s machinery, the aims of truth and of settled belief are wheels that engage one another; reflection on the methods under consideration unsettles beliefs that result from them because the methods are unlikely to lead to beliefs that agree with the facts.

7. [In this chapter, quotations of Hume are based on an electronic edition prepared by Thomas Beauchamp, David Norton, and Alexander Stewart (BNS). The chapter employs the following abbreviations, in addition to those at “Abbreviations for Editions of Seventeenth- to Nineteenth-Century Works”: CP—Collected Papers of Charles Sanders Peirce, volumes 1–6, edited by Charles Hartshorne and Paul Weiss, 1931–35, and volumes 7–8, edited by A. W. Burks, 1958, Harvard University Press, Cambridge, Massachusetts (references to volume and paragraph number); M—Sextus Empiricus, Against the Mathematicians, R. G. Bury, translator, Sextus Empiricus, volumes 2–4, Harvard University Press, Cambridge, Massachusetts, 1935, 1936, and 1939; PH—Sextus Empiricus, Outlines of Pyrrhonism, R. G. Bury, translator, Sextus Empiricus, volume 1, Harvard University Press, Cambridge, Massachusetts, 1933.]

120

reflection and the stability of belief

This interpretation disarms an objection to Peirce due to Harry Frankfurt. Frankfurt is prepared to suppose, for the sake of the objection in question, that Peirce has established that the goal of inquiry is fixed belief. Frankfurt writes: “But to propose this is to deprive inquiry of its cognitive significance, for it is to run afoul of the distinction between capricious and serious belief which has already been discussed.” The reference is to Frankfurt’s discussion of a “man who has written down on separate slips of paper all the possible answers to a question and decided to believe the answer on the slip that he picks while blindfolded.” Frankfurt observes that the man “surely was not engaged in inquiry.” The issue is not whether Peirce wants to locate a sense in which belief might be capricious. Peirce is explicit that beliefs based on authority are determined by “caprice” (CP 5.382), and that beliefs based on tenacity, authority, and the a priori method have an “accidental and capricious element” (5.383). The issue is whether Frankfurt is right to suggest that “it was necessary for [Peirce] to introduce into his discussion considerations unwarranted by his original account of inquiry.”8 Frankfurt views 5.383 as constituting a “revision” in Peirce’s conception of inquiry: “Here Peirce suggests that behind inquiry lies a desire not merely for fixed beliefs, but for beliefs that are fixed in accordance with the facts.”9 Peirce, however, is not guilty of illicitly importing truth into his account of the aim of inquiry. Peirce’s position simply incorporates into his account of the aim of inquiry, settled belief, the aim of belief itself, agreement with the facts. If Peirce took the aim of inquiry to be the settlement or fixation of blind faith—as characterized earlier in this section, where blind faith is not associated with a disposition to regulate one’s first-order dispositions by evidence of their truth—the considerations about “the facts,” about the truth of one’s convictions, would be out of place. For Peirce, it is the irritation specifically of doubt that motivates inquiry; doubt is the opposite of belief (CP 5.372), and belief is a way of regarding propositions as true that aims to coincide with the facts. Frankfurt writes that, for Peirce, “[t]he sole relevant criterion in the evaluation of any method, as such, is whether or not it is effective in securing the result for which it is designed” (emphasis added).10 This overlooks the relevance of a distinction between lower- and higher-order aims. It might be helpful to recall that Peirce introduces ‘inquiry’ as a term of art, hastening to allow that his terminology is potentially misleading: “The irritation of doubt causes a struggle to attain a state of belief. I shall term this struggle Inquiry, though it must be admitted that this is sometimes not a very apt designation” (CP 5.374). Peirce uses the term ‘inquiry’ in the sense of the process motivated by the desire for belief. The term is, I suspect, more commonly associated with the lower-order desire to regulate one’s belief that a proposition is true by evidence of its truth. 8. The quotations of Frankfurt in this paragraph are from 1958, 590, 588, 590, 591, respectively. 9. Frankfurt 1958, 592. 10. Frankfurt 1958, 590.

sextus, descartes, hume, and peirce

121

Throughout this chapter, when I write, without special explanation, of the objective of inquiry, I have in mind a higher-order objective than aiming at truth. This helps to clarify my initial statements of my position—that for Descartes and Hume, as well as Sextus and Peirce, the objective of inquiry is characterized in psychological terms. My thesis is that the four figures share the view that the desire to aim at truth that is characteristic of belief is itself subject to a higher-order desire, the desire to secure doxastic states that satisfy conditions characterized in psychological terms, without reference to truth.

2. Descartes on Doubt and Unshakability I begin with a development of Descartes’ account of the objective of inquiry. The first step is to identify an important strand in Descartes’ conception of doubt. In paragraph fourteen of Meditation V, Descartes considers circumstances in which he remembers having clearly and distinctly proved a geometrical theorem, but without attending to the proof at the time of the recollection: [O]ther arguments can now occur to me which might easily undermine my opinion, if I did not possess knowledge of God; and I should thus never have true and certain knowledge about anything, but only shifting and changeable opinions. . . . [I]n spite of still remembering that I perceived [the proof of the theorem] very clearly, I can easily fall into doubt about its truth, if I am without knowledge of God. For I can convince myself that I have a natural disposition to go wrong from time to time in matters which I think I perceive as evidently as can be. (CSM 2.48: AT 6.69–70) In context, easily having one’s opinion undermined and easily falling into doubt are alternative descriptions of what might happen to a person who lacks knowledge of the existence of a nondeceiving God. Falling into doubt is sufficient for having one’s opinion undermined. Descartes tells us that when his opinion is undermined, he is left with “shifting” or inconstant (vagas) and “changeable” (mutabiles) opinions.11 Falling into doubt, therefore, is sufficient for opinion that is shifting and changeable.12 Doubt is sufficient, we might say, for unsettled belief. A 1640 letter to Regius confirms a connection between doubt and unsettled belief. Descartes is again discussing circumstances where we recollect a 11. The translation of ‘vagas’ as ‘vague’ at HR 1.184 seems unsupportable; vagueness is nowhere at issue. This translation also appears in Cress 1979, 44, and Heffernan 1990, 179. For ‘experientia vaga’ as a technical term in Spinoza, see Curley 1985, 636. 12. Hobbes identifies doubt with a “whole chain of opinions [that] alternate in the question of true and false” (Lev. I, vii).

122

reflection and the stability of belief

conclusion that we previously clearly and distinctly perceived, but without attending to the demonstration: I say that on such occasions, if we lack knowledge of God, we can imagine that the conclusions are uncertain even though we remember that they were deduced from clear principles: because perhaps our nature is such that we go wrong even in the most evident matters. Consequently, even at the moment when we deduced them from those principles, we did not have knowledge [scientia] of them, but only a conviction [persuasio] of them. I distinguish the two as follows: there is conviction when there remains some reason which might lead us to doubt, but knowledge is conviction based on a reason so strong that it can never be shaken by any stronger reason. Nobody can have the latter unless he also has knowledge of God. (CSMK 147: AT 3.64–65) Descartes characterizes mere conviction as susceptibility to doubt. He characterizes conviction that constitutes scientific knowledge as conviction that cannot be shaken. Mere conviction and scientific knowledge would not be mutually exclusive if a conviction could be susceptible to doubt and yet unshakable. The passage presupposes that susceptibility to doubt is sufficient for shakability.13 Doubt leaves our previous beliefs shaken. ‘Shaken’, in Kenny’s translation of the letter to Regius, renders ‘concuti’ (from ‘concutere’) to shake or agitate. This is another way of saying the belief is unsettled by doubt. It is worth noting that twice in the Meditations, Descartes writes of seeking something that is “unshakeable,” rendering ‘inconcussum’ (CSM 2.16, 17: AT 7.24, 25).14 The letter to Regius and paragraph fourteen of Meditation V offer accounts of the requirements for scientific knowledge (scientia) and for “true and certain knowledge,” respectively. I take these pieces of terminology to belong to a cluster of interchangeable technical terms. In both the letter to Regius and Meditation V, scientific knowledge is belief that is incapable of being unsettled or shaken.15 And in both passages, doubt shakes or unsettles belief. These passages enable us to give substance to the truism that

13. Perhaps more strictly, the upshot of the letter is that doubt is at least able to shake or unsettle belief; this falls short of the result of the Meditation V passage, that doubt is sufficient for unsettled opinion. It is unclear, however, what beyond doubt would be required to unsettle opinion. I see no reason to think that in the letter to Regius Descartes seeks to weaken his position in Meditation V. 14. The CSM translation as ‘unshakeable’ is strictly speaking a stretch from the more literal ‘unshaken’, though perhaps supported by the use of ‘indubitable’ (AT 9A.19, 20) in the authorized French translation. I am grateful to David Velleman for calling my attention to the issue about translation, and to Edwin Curley for help in addressing it. What matters most for my purposes is Descartes’ use of a cognate of ‘concutere’ in the Meditations. 15. For additional discussion, and for the place in the literature of my interpretation of Descartes’ conception of unshakability, see my 1992a [this volume, ch. 3], esp. §§1–2.

sextus, descartes, hume, and peirce

123

Descartes seeks belief that is certain or indubitable, as well as to give sense to his persistent metaphors of firm and solid belief (CSM 1.115, 126, 2.12, 103, 104: AT 6.9, 31, 7.17, 145, 146): Descartes seeks beliefs that are incapable of being shaken.16 Descartes’ objective of belief that is incapable of being unsettled is similar to Peirce’s objective of the “settlement” of opinion, or “fixation” of belief, as Peirce’s understanding of the objective is standardly interpreted. (Indeed, Peirce uses “shake,” “shaken,” and “unshakable” in related contexts—see CP 5.378, 2.29, 5.516). On this interpretation, the fixation of belief requires belief that is settled permanently or in the long run.17 This looks roughly equivalent to the Cartesian objective of unshakability. It would seem that permanence in belief (at least, permanence that does not arise by accident) would require unshakability, and that belief that is incapable of being unsettled would be permanent.18 We have seen that Descartes holds that doubt is sufficient for unsettled belief. My formulation has been intentionally ambiguous. It could mean that doubt is causally sufficient to unsettle belief, or it could mean that Descartes identifies doubt with an unsettled state (and belief with a settled state). We have just this identification in Peirce. For Peirce, doubt is a hesitating, wavering, unsettled state; belief is opposed to doubt, so that belief is a state that is firm or settled (CP 5.372, 394, 417). The reading of Descartes which identifies doubt with an unsettled state is the most economical and captures the direction or tendency of his thought in passages we have considered. (The economical reading, however, is perhaps somewhat anachronistic. In the official ontology, doubt is a conscious experience, a mode of thought.) Whether or not he identifies doubt with an unsettled state, it remains the case that for Descartes doubt is unsettling and that the objective of inquiry is a settled state that is incapable of being unsettled. Peirce is notoriously critical of Descartes, and this might seem at odds with my interpretation of Descartes’ account of doubt. According to Peirce, doubt must be “real”—rather than “paper” or “verbal” (CP 5.376, 416, 445, 451)—in that it must actually unsettle belief, and Cartesian doubt fails this test. Real doubt requires a “positive reason” (5.265); it must be “real and living” (cf. 2.192, 5.376, 6.498); it must be particular rather than universal (cf. 5.265, 318, 376, 416); it must have an “external origin” (5.443); and it must have the capacity to change belief (cf. 5.265). Peirce’s conception of doubt is a complex topic. The gist of my position is this. Peirce’s chief quarrel is with Descartes’ view of the kinds of hypotheses

16. In the first sentence of Meditation I, where Descartes writes of establishing something firm in the sciences, the verb is ‘stabilio’, to establish or to make stable. (The verb, however, is ‘établir’ in the French—AT 9A.13). 17. Goudge 1950, 18; Murphey 1961/1993, 163; and Misak 1991, 46–47, 80–81. 18. Details of Descartes’ notion of unshakability require qualification of the latter claim. See my 1992a, esp. §§5–9. In 1990, esp. 15–16 [this volume, ch. 2, §3], I mistakenly took permanence in belief to be the central notion in Descartes’ conception of the objective of inquiry.

124

reflection and the stability of belief

that are capable of generating doubt. At the same time, Peirce’s conception of the nature of doubt was much closer to that of Descartes than Peirce thought.19 Peirce was more than capable of overlooking detail and nuance in Descartes’ philosophy. For example, as Frankfurt has shown, Peirce is mistaken in his claim: “The distinction between an idea seeming clear and really being so, never occurred to [Descartes]” (CP 5.391).20 In the present context, Peirce suggests that “no one who follows the Cartesian method will ever be satisfied until he has formally recovered all those beliefs which in form he has given up” (5.265). This is not an accurate account of Descartes’ conception of the outcome of the method of doubt. A number of beliefs about the material world with which the doubter might begin—that there is a void, and that bodies possess qualities exactly similar to sensory experiences of secondary qualities—do not survive the doubt.21 A more fundamental point is that Peirce is blind to the strand in Descartes’ conception of doubt, on which doubt is sufficient for unsettled belief, that I have identified. There is much evidence Peirce overlooks, beyond the crucial passages from Meditation V, the letter to Regius, and elsewhere. It is the clear intent of the final two paragraphs of Meditation I that its skeptical hypotheses eventually undermine opinion; noticing that his “habitual opinions keep coming back,” Descartes proposes a technique for overcoming “the distorting influence of habit” (CSM 2.15: AT 7.22). By the first paragraph of Meditation II, the doubt Descartes has introduced is unsettling the meditator: “So serious are the doubts into which I have been thrown as a result of yesterday’s meditation that I can neither put them out of my mind nor see any way of resolving them. It feels as if I have fallen unexpectedly into a deep whirlpool which tumbles me around” (CSM 2.16: AT 7.23–24). It is at the close of this paragraph that Descartes notes that “Archimedes used to demand just one firm and immovable point in order to shift the entire earth” and expresses his interest in managing “to find just one thing, however slight, that is certain and unshakeable” (CSM 2.16: AT 7.24). In addition, Descartes holds that a proposition cannot be doubted at any time it is clearly and distinctly perceived (CSM 2.104, 309, 321: AT 7.146, 460, 477). Indeed, Descartes maintains that a proposition is not “in any way” (nullo modo) open to doubt (AT 7.38, 8A.21: CSM 2.27; cf. 1.207)

19. I do not seek to minimize the differences between the conceptions of doubt in Peirce and Descartes. Peirce holds that belief is inherently connected to action (cf. CP 5.371, 373, 397–98). Belief is settled because it is habitual, and a habit leads a person to act in particular ways in particular sorts of circumstances. Descartes, however, divorces action from belief (CSM 2.15; CSMK 229: AT 7.22, 4.63). We can speculate that this misleads Peirce into thinking that for Descartes doubt need not be psychologically real. I introduce another difference at the close of §3. 20. See Frankfurt 1970, ch. 13. 21. I develop this point in my 1990 [this volume, ch. 2], §2. For misgivings in regard to Peirce’s understanding of Descartes on doubt, see Haack 1982, 162–67, and Hookway 1990b, 398.

sextus, descartes, hume, and peirce

125

when it is clearly and distinctly perceived.22 It is difficult to see how Descartes could claim this if doubt, on his conception of it, need not be psychologically living or real. Granted, Descartes also holds that belief in a proposition is psychologically irresistible at any time it is clearly and distinctly perceived. This doctrine appears in the letter to Regius—“our mind is of such a nature that it cannot help assenting to what it clearly understands” (CSMK 147: AT 3.64)—in paragraph fourteen of Meditation V (CSM 2.48: AT 7.69), and elsewhere (cf. CSM 1.197, 207, 2.103, 104; C 6: AT 8A.9, 21, 7.145, 146, 5.148).23 But if Peirce is correct that Cartesian doubt can be merely paper or verbal, on what grounds could Descartes claim that it is not possible, while irresistibly believing that p, to doubt whether p is true?24 The explanation cannot be that belief and doubt are incompatible or opposite states. Perhaps they are, but if so, belief would preclude doubt whether the belief is irresistible or not. If doubt is a state that unsettles belief, Descartes seems entitled to the claim that irresistible belief is not “in any way” open to doubt. We need only suppose that, in Descartes’ view, irresistible belief cannot be unsettled. Obviously, the clear and distinct perception of a proposition can be disrupted. Descartes is thus careful to claim that a belief is irresistible and not subject to doubt “so long as,” or “as long as,” it is clearly and distinctly perceived (CSM 1.197, 2.48, 309; C 6: AT 8A.9, 7.69–70, 460, 5.149). Given that belief cannot be unsettled so long as it is irresistible, it is because doubt is a state that unsettles belief that it follows that irresistible belief precludes doubt.

3. The Role of Unshakability in Descartes’ Epistemology What is the function, in Descartes’ epistemology, of the strand in his thinking that identifies unshakability as the objective of inquiry? The answer is that the conceptions of doubt and unshakability discussed in the preceding

22. CSM does not capture the identity in the Latin expressions. They have “cannot in any way be open to doubt” for the first of the two passages and “quite unable to doubt” for the second. The more literal Miller and Miller translation of the passage from the Principles seems superior (1983, 20). 23. At paragraph fourteen of Meditation V, Descartes extends this doctrine to demonstrations or proofs, connected sequences of clear and distinct perceptions, as well as to individual clear and distinct perceptions (cf. CSM 1.197; C 6: AT 8A.9, 5.148). 24. The room for this question can be obscured by running together the claims that clear and distinct perception is indubitable and that it is irresistible, as in Frankfurt: “But [Descartes] also holds that he cannot doubt the truth of what he perceives while he is perceiving it clearly and distinctly. ‘Our mind is of such a nature’, he says, ‘that it cannot refuse to assent to what it apprehends clearly’. Descartes enunciates this doctrine on a number of occasions, but never explains his grounds for it” (1970, 163). Here Frankfurt treats irresistibility as equivalent to indubitability.

126

reflection and the stability of belief

section are essential elements in Descartes’ solution to the problem of the Cartesian circle. In a nutshell, the problem of the circle arises as follows. In the final four paragraphs of Meditation I, Descartes invokes the hypothesis of a powerful deceiver, raising doubt about the truth of, at least, beliefs based on sense-perception.25 In paragraph four of Meditation III, Descartes also invokes a hypothesis of a powerful deceiver, in this instance raising doubt specifically about the truth of beliefs based on clear and distinct perception. In response to these skeptical hypotheses, Descartes offers proofs of the existence of an all-perfect being, God. Since an all-perfect being is omnipotent and a nondeceiver, God would not give us cognitive faculties— such as the faculties of sense-perception and clear and distinct perception— which lead to false belief if used correctly. The proofs of the existence of God, however, rely on clear and distinct perception, so that Descartes’ response to the deceiver hypothesis appears question-begging. Descartes took the problem of the circle seriously and provided an answer. His approach might be called a psychological response to the problem of the circle, a line of response that has now received considerable attention in the literature.26 The response relies on Descartes’ claim that doubt unsettles belief, so that the skeptical hypotheses unsettle the beliefs whose truth they call into question. Drawing on the framework introduced in §1, we can state the solution as follows. While belief aims at truth, belief is directed by a higher-order objective, unshakability. Descartes thinks that unshakability can be achieved, in light of two theses identified in §2: that belief cannot be shaken so long as it is psychologically irresistible and that belief is psychologically irresistible so long as one attends to a clear and distinct demonstration of the proposition believed. The objective of unshakability can then be achieved by someone who possesses a clear and distinct demonstration of the existence of a nondeceiving God and its epistemological implications. This demonstration does beg the question with respect to the truth of the skeptical hypothesis of Meditation III. At the same time, when a person attends to the relevant demonstration, he irresistibly believes that there is a nondeceiving God, that his cognitive faculties would not lead to false belief if used correctly, that whatever he clearly and distinctly perceives is true, and so forth. Such beliefs are unshakable; the skeptical hypothesis cannot unsettle them, so long as they are clearly and distinctly perceived. What has happened here to aiming for truth? Descartes does not claim that he cannot achieve truth. The deceiver hypothesis has no tendency to show that our cognitive faculties are unreliable, in the sense of leading to false beliefs more often than not; the hypothesis at most shows that our cognitive faculties might be unreliable. True belief remains an objective. 25. Whether the deceiver hypothesis in Meditation I calls into question beliefs based on clear and distinct perception, as well as beliefs based on sense-perception, has been a matter of controversy. For the central discussion of the matter, see Frankfurt 1970, part 1, 3–87 26. See Rubin 1977; Larmore 1984; and Bennett 1990. What follows is a sketch of the psychological response. For required refinements, see my 1990, §6, and 1992a, esp. §§5–9.

sextus, descartes, hume, and peirce

127

Short of begging the question, however, Descartes fails to locate a way of establishing that clear and distinct perception and other cognitive faculties lead to truth. Truth is an objective that gives out, in the sense that Descartes is unable to exploit considerations related to truth to ground a particular method for forming beliefs—such as reliance on clear and distinct perception—without begging the question. More generally, an aim “gives out” if it underdetermines the choice among means, in light of the information available, thereby allowing a higher-order aim to determine the choice. Suppose I undertake to play a grand master. Were I to win, it would be a matter of sheer luck; no strategy is better suited to my winning than any other, so that the aim of winning gives out. If I am playing with the higher-order objective of having fun, I might as well regulate my style of play accordingly, among the various ways I might try to win. Similarly, Descartes does think he is able to exploit the higher-order objective of unshakability, coupled with some psychological theses, to ground a method for forming belief. The higherorder objective of unshakability takes over where the lower-order objective gives out.27 For Descartes, the aim of truth remains in place, though it is idle in his response to the problem of the circle.28 Here we have a difference between Descartes and Peirce. As we have seen in §1, in Peirce’s assessments of methods of forming beliefs, the aims of truth and unshakability engage each other. Reflection on tenacity, authority, and the a priori method is unsettling because there is no reason why beliefs based on these methods should coincide with the facts; by contrast, “to coincide with the fact . . . is the prerogative of the method of science” (CP 5.387). Peirce’s assessments of belief-forming methods rely on considerations about the likely reliability of a method in forming true beliefs. At the stage of the Meditations where Descartes must respond to the hypothesis of a deceiver, such considerations are out of bounds, in that they would beg the question with respect to the truth of the skeptical hypotheses. It is the Pyrrhonian, not Descartes, who gives up the aim of truth. Lowerand higher-order aims can conflict. The conception of a hierarchy of desires 27. These observations constitute my response to A. Miller’s claim that there is a “crucial mistake” (1994, 118) in the argument for the infallibility of reason that I attribute to Descartes in my 1990. Miller’s objection rests on the thought that the “objectives of truth and permanence stand side by side” (119). (“Permanence” is the ancestor of “unshakability” in my interpretation here.) Truth and unshakability cannot, for Descartes, stand “side by side”; the skeptical hypotheses preclude Descartes, during the relevant stages of his argument, from appealing to considerations of truth. This response was implicit in my 1990, §6. 28. Bennett comes very near to my position: “Descartes has some concern with stability considered as standing on it [sic] own feet and not as an upshot of truth” (1990, 76), so that there is a strand in Descartes where stability is “considered as a goal to be reached directly rather than through truth” (80). Bennett goes too far, however, in writing: “To the extent that his goal is just stability, Descartes can tell a coherent story about what his belief in God’s veracity does for him” (102). Descartes’ goals are truth and stability, not “just stability,” but the goal of truth does no work in Descartes’ solution to the problem of the circle. Similarly, we should not construe Descartes as “setting the pursuit of truth aside” (105), except in the sense that the pursuit of truth does not ground the adoption of a particular method.

128

reflection and the stability of belief

carries no implications about the relative strengths of the higher- and lowerorder desires. The higher-order desire, however, might be the stronger, so that in cases of conflict the higher-order aim would take precedence. For example, one might be so inept at playing chess that one simply cannot have fun playing. If having fun takes precedence, one would give up playing chess in favor of some other way of having fun; the higher-order aim of having fun would supplant the lower-order aim of winning at chess. Pyrrhonian skepticism is an example of a position on which a higher-order desire for a psychological objective supplants the lower-order desire for truth. Though the skeptic sets out to achieve true belief as a means to quietude (see §1), he “found himself involved in contradictions of equal weight, and being unable to decide between them suspended judgment” (PH I.26). The skeptic gives up achieving quietude by aiming for true belief. Sextus continues: [T]he Skeptics were in hopes of gaining quietude by means of a decision regarding the disparity of the objects of sense and of thought, and being unable to effect this they suspended judgment; and they found that quietude, as if by chance, followed upon their suspense. (I.28–29) For the Pyrrhonian, the lower-order aim of truth is renounced as one that cannot be achieved, and the higher-order objective of quietude is achieved in another way, by suspending belief. Descartes, however, does not claim that there is an incompatibility either between aiming at truth and aiming at unshakability or between achieving truth and achieving unshakability; for Descartes, the higher-order goal of unshakability does not supplant the lowerorder goal of truth. I do not claim that Descartes’ solution to the problem of the circle is a satisfying one.29 Where the aim of truth gives out, the aim of unshakability takes over, faute de mieux. Descartes’ position amounts to his settling for showing how to achieve unshakability, absent a showing (without begging the question) that unshakable beliefs are likely to be true.30 This raises the possibility of a developmental interpretation of Descartes, according to which he adopted the higher-order aim of unshakability under pressure from the problem of the Cartesian circle. Descartes’ emphasis on the unsettling character of doubt and on unshakability as an objective are prominent in texts where Descartes is discussing the problem of the circle—Meditation V, the Replies, and the letter to Regius. On the other hand, some of the evidence presented in §2 for Descartes’ conception of doubt derives from Meditations I and II. My own speculation is that Descartes started out with a conception of an objective of “firm” belief, thinking of such belief as 29. I have suggested elsewhere that Descartes can do somewhat better putting considerations of truth in play than in the position described here. See my 1992a, §10. 30. As Bennett writes, Descartes “is capable, at least sometimes, of . . . settling for something subjective, psychological, causal—something like stability” (1990, 105).

sextus, descartes, hume, and peirce

129

psychologically unshakable because based on the best possible evidence of truth. If so, the aim of unshakability is in place from the beginning, with Descartes expecting unshakability to be a by-product of evidence of truth. Unable to identify such beliefs without begging the question against the deceiver hypothesis, he becomes prepared to rely on unshakability on its own.31 What matters more for my purposes is that Descartes offers little independent argument for adopting unshakability (that is, mere psychological unshakability) as a higher-order objective. Perhaps he comes closest to doing so in the Second Replies, where Descartes supplies a forceful statement of unshakability as an objective of inquiry. The statement leads to an account of “perfect certainty,” which I take to belong to the cluster of terms interchangeable with ‘scientific knowledge’ (§2). In the sentence following the quotation I am about to give, Descartes identifies perfect certainty with “firm and immutable conviction” (CSM 2.103: AT 7.145). Descartes writes: First of all, as soon as we think that we correctly perceive something, we are spontaneously convinced that it is true. Now if this conviction is so firm that it is impossible for us ever to have any reason [causa] for doubting what we are convinced of, then there are no further questions for us to ask: we have everything that we could reasonably want. What is it to us that someone may make out that the perception whose truth we are so firmly convinced of may appear false to God or an angel, so that it is, absolutely speaking, false? Why should this alleged ‘absolute falsity’ bother us, since we neither believe in it nor have even the smallest suspicion of it? For the supposition which we are making here is of a conviction so firm that it is quite incapable of being destroyed; and such a conviction is clearly the same as the most perfect certainty. (CSM 2.103: AT 7.144–45) This is reminiscent of Peirce, in “The Fixation of Belief”: [T]he sole object of inquiry is the settlement of opinion. We may fancy that this is not enough for us, and that we seek, not merely an opinion, but a true opinion. But put this fancy to the test, and it proves groundless; for as soon as a firm belief is reached we are entirely satisfied, whether the belief be true or false. (CP 5.375; cf. 376, 563, 6.485) Similarly, Peirce wrote three decades later: “If you absolutely cannot doubt a proposition—cannot bring yourself, upon deliberation, to entertain the 31. This view is similar to that of Bennett 1990, esp. 75–76, 102–5, 107–8, though Bennett’s position incorporates hypotheses about aspects of Descartes’ thinking that were subliminal (cf. 107). Adjudicating the questions of the development of Descartes’ view would require a detailed inspection of the texts, and hypotheses about the composition of the Meditations, that go well beyond the purposes of this chapter.

130

reflection and the stability of belief

least suspicion of the truth of it, it is plain there is no room to desire anything more” (6.498). For Peirce, when we reach firm belief, “we are entirely satisfied,” and have everything we could “desire,” whether the belief be true or false. For Descartes, if we achieve conviction so firm that it cannot be destroyed, “we have everything that we could reasonably want,” even if the belief is alleged absolutely false.32 But why do we want unshakable conviction at all? Descartes does not say. As I have suggested, this objective comes to the forefront under pressure of the problem of the circle; Descartes does not provide an explicit independent rationale for it. Peirce, by contrast, provides a rationale: doubt is an uneasy and dissatisfied state, and the irritation of doubt provides the only immediate motive for inquiry, the struggle to attain belief. In providing resources for an account of the selection of the objective of inquiry, Peirce supplies a layer of theory that is virtually absent in Descartes.

4. Peirce and Hume on Doubt In the claim that doubt is an uneasy and dissatisfied state, Hume, unlike Descartes, is aligned with Peirce. In the final paragraph of Treatise II.iii.10, “Of curiosity, or the love of truth,” Hume makes a number of important claims about uncertainty or doubt, and certainty or belief (T 453). In the first place, there are claims about the “nature” of doubt and belief. Hume writes: “’tis the nature of doubt to cause a variation in the thought,” so that doubt involves “instability and inconstancy” (453). By contrast, belief or certainty involves “fixing one particular idea in the mind, and keeping it from wavering in the choice of its objects” (453). In this passage, Hume treats belief and certainty as the opposites of doubt and uncertainty. If it is of the nature of doubt to cause “variation,” “instability,” and “inconstancy” in thought, then the property of “fixing one particular idea in the mind, and keeping it from wavering,” must be of the nature of belief.33 It is in the nature of belief, unlike doubt, to be settled.34 In the second place, there are claims about contingent psychological effects of doubt and belief, as distinct from claims about their nature. Hume

32. Frankfurt brought the Second Replies passage into prominence with the suggestion that it commits Descartes to a coherence theory of truth (1970, 179; cf. 25–26, 170). Frankfurt would have done better to jettison this aspect of his interpretation, falling back on his emphasis on Descartes’ interest in belief that is solid, permanent, and unshakable (179–80; cf. 24, 44, 45, 124). 33. Hume writes at I.iii.7, “Of the nature of the idea or belief,” that “’till there appears some principle, which fixes one of these different situations, we have in reality no opinion” (T 96). 34. These claims might seem at odds with Hume’s characterizations of belief with reference to vivacity, liveliness, or intensity. To remove the apparent conflict, it is necessary to exploit Hume’s tendency to treat belief as a disposition, as well as an occurrent state. I take up these matters in my 1995a [this volume, ch. 7], §4, and 1995b, §2.

sextus, descartes, hume, and peirce

131

writes in the preceding section of the Treatise that “uncertainty . . . is uneasy” (T 447).35 Hume tells us at Selby-Bigge page 453: “’Tis a quality of human nature, which is conspicuous on many occasions, and is common both to the mind and body, that too sudden and violent a change is unpleasant to us, and that however any objects may in themselves be indifferent, yet their alteration gives uneasiness.” Since “’tis the nature of doubt to cause a variation in the thought, and transport us suddenly from one idea to another, it must of consequence be the occasion of pain” (453).36 Doubt is uneasy because it is unsettled. By contrast, belief or certainty “prevents uneasiness” (453). Hume holds independently that there is a general desire for pleasure, and aversion to pain (cf. 118–19, 414, 438–39, 574–75). The uneasiness in doubt, therefore, provides a motive for its own removal. There is a natural motive to replace doubt in favor of belief. The central claim at Treatise 453 recurs in the Appendix. In belief, unlike doubt, the mind “fixes and reposes itself in one settled conclusion” (T 625); whereas doubt is associated with “agitation,” and hence uneasiness, belief is pleasant, involving “tranquility and repose” (626). Hume’s position in the passages at pages 453 and 625–26 bears a striking similarity to that of Peirce in “The Fixation of Belief”: “Doubt is an uneasy and dissatisfied state from which we struggle to free ourselves and pass into the state of belief; while the latter is a calm and satisfactory state which we do not wish to avoid” (CP 5.372). For Peirce, doubt is a state of uneasiness or dissatisfaction that provides an immediate motive to seek relief in favor of the tranquil or calm state where thought is settled, or does not fluctuate (cf. 2.210, 5.384, 394, 397, 510, 605). Hume adopts a similar position in Book II and the Appendix.37 Furthermore, it is explicit in Hume, and implicit in Peirce, that doubt is unpleasant because it is unsettled. We should expect Hume’s epistemological position to emerge in Book I of the Treatise, so that the evidence from Book II and the Appendix cannot on its own sustain an interpretation of his epistemology.38 The Peircean idea that we struggle to remove the uneasiness in uncertainty or doubt is prominent,

35. Hume says: “uncertainty alone is uneasy.” In context, “alone” means on its own, not only. 36. MacNabb is the only commentator I know to take note of the fact that Hume takes instability to be unpleasant, something we dislike (1951/1966, 99, 192). Curiously, he does not bother to cite any textual evidence for this. 37. At Treatise 453, Hume is explaining the source of “curiosity.” He provides a distinct, and more extended discussion, of the “love of truth” at 448–52, a discussion that arguably seeks to explain the love of inquiry. I take this point. Peirce, however, is providing an account of the immediate motive for inquiry, or the struggle to attain belief. Hume can be a Peircean in this regard without supposing that the relief of the uneasiness in doubt is the only motive to inquiry. 38. Though they do put forward the claim that doubt is uneasy as a general doctrine, the discussions at II.iii.10 and the Appendix emerge for rather specialized purposes. In the Appendix, Hume is defending the claim that belief is a modification of an idea, not an impression distinct from an idea.

132

reflection and the stability of belief

however, in I.iv.2, “Of skepticism with regard to the senses.” According to Hume, there is a propensity to ascribe identity to a succession of perceptions—such as the perceptions of a mountain, or of furniture in the chamber, before and after we shut our eyes or turn our head (T 194–95)—that are perfectly resembling or invariable, though interrupted (204). At the same time, the interruption in the perceptions makes us consider them distinct objects (205). We are involved in a “contradiction” (199, 205, 208), and hence “perplexity” (205), that is, uncertainty or doubt. In these circumstances, there is an “opposition” (206), a “combat of internal principles” (205). The metaphors point to a psychological war, a highly unsettled condition. This is a source of uneasiness. Hume writes: [A]ny contradiction either to the sentiments or passions gives a sensible uneasiness. . . . Now there being here an opposition betwixt the notion of the identity of resembling perceptions, and the interruption of their appearance, the mind must be uneasy in that situation, and will naturally seek relief from the uneasiness. Since the uneasiness arises from the opposition of two contrary principles, it must look for relief by sacrificing the one to the other. (205–6, emphasis added) In this we are frustrated, since the opposing beliefs prove equally strong (cf. 206). In the end, we relieve our uneasiness by supposing that perfectly resembling perceptions have a continued existence at times when they are not perceived, and hence an unchanging and uninterrupted existence (cf. T 199, 205–10). This supposition leads to a second contradiction. Hume thinks that reflection on the phenomena of double vision and perceptual relativity convinces us that perceptions do not have an existence independent of the mind, and hence do not have a continued existence when not perceived (cf. 210, 214, 215). This conflicts with the supposition at hand. In connection with this “contradiction” (215), Hume identifies the same elements of doubt, struggle, and uneasiness encountered in the previous example. We are not “fully convinc’d” (215, twice) of either side of the contradiction, so that our mind is in an “intermediate situation” (216), that is, doubt or uncertainty. There is “opposition” between “two enemies” that “struggle” to “destroy” one another (215). Hume writes: “In order to set ourselves at ease in this particular, we contrive a new hypothesis” (215, emphasis added); “Not being able to reconcile these two enemies, we endeavour to set ourselves at ease as much as possible, by successively granting to each whatever it demands” (215, emphasis added). We are led, that is, to the hypothesis of the double existence of perceptions and objects, to indirect or representative realism. Treatise I.iv.2 provides extensive confirmation of the Peircean themes of page 453 and the Appendix: doubt is an unsettled and hence uneasy state, which we desire to relieve.39 Here we have a higher-order desire than aiming 39. I have more to say about connections between Hume and Peirce in my 1995b.

sextus, descartes, hume, and peirce

133

at truth—the desire to attain belief for the sake of relieving uneasiness and achieving tranquility—so that the objective of inquiry is settled belief.40 On Hume’s view, one ought to seek doxastic states that are settled, or in equilibrium.41 One ought to do so in order to relieve the uneasiness in an unsettled state; uneasiness provides a natural motive to relieve doubt in favor of belief. For Hume, epistemic obligation is naturalized, as deriving from the motivational force of the felt uneasiness to which an unsettled doxastic condition gives rise.42 Though my interpretation of Hume’s epistemology is “naturalistic,” it is importantly different from that due to Norman Kemp Smith. Kemp Smith maintains that Hume has a response to skepticism in his doctrine of “natural beliefs” that are “inevitable” or “irresistible.”43 This leads Kemp Smith to formulate a sense in which, for Hume, natural beliefs ought to be accepted: “The beliefs which ought to be accepted are, [Hume] teaches, beliefs that Nature itself marks out for us. In their fundamental forms, as ‘natural’ beliefs, we have no choice but to accept them; they impose themselves upon the mind.”44 On the Kemp Smith interpretation, Hume’s epistemology is entirely negative: since the natural beliefs are irresistible, to say that we ought not hold them is pointless, or even false (if ought implies can).45 On my interpretation, Hume not only delimits the class of beliefs we ought to accept in naturalistic terms but also provides a positive account of our obligation to accept them with reference to the desire to relieve the uneasiness in unsettled states. What is more, as I interpret Hume, irresistibility is neither necessary nor sufficient for justification: beliefs could be settled without being irresistible; and conflicting beliefs could be irresistible without being settled.

40. In my 1990, §7, I mistakenly took permanence in belief, rather than settled belief, to be the central notion in Hume’s conception of the objective of inquiry. 41. A number of commentators have called attention to the importance of a notion of stability in Hume’s theory of belief and epistemology—most notably MacNabb 1951/1966, 72–79, 96–100 (cf. 166–67, 191–93), and, more recently, Fogelin 1985, 60–62, 75, 83, 92, and Baier 1991, 5–6, 16, 24, 58, 72–74. For the most part, these commentators offer remarks about stability somewhat in passing, or within the confines of particular epistemological discussions. They do not advance a systematic stability-based interpretation of Hume’s theory of justification. (Baier’s comments at 26–27 undercut her recognition of the importance of stability for Hume: a Humean “reformed philosopher makes no bogey out of contradiction”; “There will, in the nature of the enterprise, be contradictions between various parts of the Treatise,” giving rise to “delicate dialectical satisfactions.” Here it looks as if contradiction and instability are to be prized.) 42. I elaborate the interpretation in my 1991, 1995a, and 1995b. 43. Kemp Smith 1941, 87, 455, 486, and 1905, esp. 152, 161, 162. 44. Kemp Smith 1941, 388 (cf. 46, 68), and 1905, 152. 45. For statements of the Kemp Smith interpretation, as characterized here, see Lenz 1958, 559, 566–67; P. F. Strawson 1958, 20–21, and 1985, 10, 11; and Stroud 1977, 76, 247 (cf. 248).

134

reflection and the stability of belief

5. Hume and Sextus Hume is interested in equilibrium or a settled condition in belief because he was working within a philosophical tradition that emphasizes the desirability of ataraxia, a state of quietude, in which one is tranquil, or not disturbed.46 A hallmark of Pyrrhonian skepticism is the claim that the suspension of belief is the only route to tranquility. Isostheneia, an equipollence of opposing arguments, leads to epoché, suspense of judgment (PH I.8), which in turn leads to ataraxia (I.26, 29, 31). Hume is concerned with tranquility in his theory of belief (T 119, 626), as well as in his theory of the passions (417, 423, 437, 442). Hume shares a Pyrrhonian conception of tranquility as the objective of inquiry. At the same time, Hume has two criticisms of Pyrrhonian skepticism. One criticism is well-known: there are beliefs—the belief in body, for example—that cannot be suspended, so that the Pyrrhonian prescription is to no avail in the effort to achieve tranquility.47 This is a criticism of the Pyrrhonian technique for achieving tranquility, not a criticism of the Pyrrhonian conception of the objective itself. Formulating the second of the two criticisms requires additional background. Such notions as “ataraxia,” “quietude,” or “tranquility” admit of more than one meaning. On the one hand, they can refer to a feeling, an occurrent state—either a calm feeling, or the absence of any felt disturbance or irritation. For Sextus, ataraxia had this meaning at least, for he took the fact that disturbances such as cold or thirst are unavoidable to imply that ataraxia cannot be fully achieved. For this reason, the objective of skepticism is ataraxia in matters of opinion and moderation of those disturbances that are unavoidable (PH I.25). Such moderation can be achieved, if we eliminate the opinion that the disturbances are good or bad (cf. I.29–30, and M XI.147–50).48 In addition to referring to a calm feeling, such notions as “tranquility” can refer to the settled character of the mental disposition that a calm feeling manifests. This is also an element in the Pyrrhonian notion: “‘Suspense’ is a state of mental rest owing to which we neither deny nor affirm anything” (PH I.10); “the term ‘suspension’ is derived from the fact of the mind being held up or ‘suspended’ so that it neither affirms nor denies anything owing to the equipollence of the matters in question” (I.196). The underlying idea is that equipollence of opposing arguments constitutes an equilibrium, and hence a settled condition, so that, other things being equal, sustained suspension of judgment, and the felt tranquility that derives from it, are 46. Immerwahr 1992 has discussed the importance of a notion of tranquility outside Hume’s epistemology. 47. Cf. Popkin 1951b, 386, 405–6; Norton 1982, 264–69; Burnyeat 1980, 20–23, and 1984, 249; Penelhum 1983a, 298, and 1983b, 35, 120–31; Fogelin 1985, 92; and Hookway 1990a, 3–4. 48. I owe the reference in the text to M to Stough 1969, 14.

sextus, descartes, hume, and peirce

135

possible.49 It is because it constitutes an equilibrium that isostheneia can lead to epoché, and in turn to ataraxia. These latter states—suspense of judgment and tranquility—require equilibrium, a settled condition that, for Sextus, is ultimately the contribution of the equipollence of opposing arguments. Hume takes a keen interest in the idea that equipollence gives rise to a feeling of tranquility and thinks that the Pyrrhonian has something right, and something wrong. Hume in effect isolates two elements in the idea. One element is that a settled condition gives rise to a calm or tranquil feeling (and that an unsettled condition gives rise to uneasiness). Hume thinks this is correct. Settled conditions in our doxastic lives, much as settled conditions in our emotional lives, feel calm. The calm feeling (or the absence of felt disturbance) just is an occurrent manifestation of the settled doxastic condition, or our conscious awareness of that state. The second element in the Pyrrhonian position is that equipollence of opposing arguments constitutes a settled condition. It is here, according to Hume, that the Pyrrhonian goes wrong. Hume’s account of why this is so constitutes the second of his two criticisms of the Pyrrhonian. We might imagine that opposing arguments somehow cancel each other out, or that they are locked in an ongoing standoff for influence on belief or assent. Hume maintains, to the contrary, that opposing arguments more typically struggle with each other in a psychological war. As we have seen in §4, the metaphors of combat are Hume’s. In I.iv.2, we have “combat,” “opposition,” and “struggle” between “enemies” that seek to “destroy” each other. Though we “look for relief by sacrificing the one to the other,” the opposing beliefs can prove equally strong (cf. T 206), so that neither “will . . . quit the field” (215). It is helpful to think in terms of Hume’s own analogy to physical combat. Equally matched combatants are not typically locked in a literal steady state. There is an ebb and flow where one combatant, then the other, temporarily gains the upper hand. Owing to the fundamental equality of the opponents, there is indeed a kind of tie or deadlock, in that neither party can prevail for long; this deadlock, however, manifests itself in an alternating cycle. Applying this model to opposing beliefs, as Hume does, we do not have a picture of an equilibrium, or settled condition, in doxastic states. We have a picture of one of the opposing beliefs ascending, and then the other.50 Thus, Hume writes in conjunction with one of the oppositions in I.iv.2 of our “successively granting to each [enemy] whatever it

49. Burnyeat writes: “if tranquillity is to be achieved, at some stage the skeptic’s questing thoughts must come to a state of rest or equilibrium” (1980, 52; and see 148n.54). Barnes translates ‘epoché’ as ‘standstill’ and writes that epoché “supervenes” where “two sets of arguments exactly balance one another” (1982, 1). Patrick 1899, 27, and Naess 1968, 5 take note that stability or equilibrium is involved in epoché and ataraxia, respectively. 50. Baier has nicely called attention to Hume’s conception of opposition in belief as dynamic, manifested in alternation over time (1982, 644–47, and 1991, 15–17).

136

reflection and the stability of belief

demands.” Similarly, in I.iv.7, in response to an opposition between two “directly contrary” operations of the mind we “successively assent to both” (T 266). Hume is explicit that in a number of instances opposing beliefs have oscillation in belief, alternating cycles of assent, as a (provisional or permanent) result.51 In Hume’s view, the second element of the Pyrrhonian position, the idea that we can achieve equilibrium in the presence of opposing arguments of equal strength, is mistaken. The presence of such arguments leads to combat, struggle, and cycles of alternating belief—a dramatically unsettled doxastic condition. Coupled with the claim that an unsettled condition gives rise to uneasiness, equipollence would be an unpleasant and disturbing state.52 Though correct that genuine mental equilibrium gives rise to felt tranquility, the Pyrrhonian has a hopeless account of how doxastic equilibrium can be achieved. This is a deeper objection to the Pyrrhonian than Hume’s first criticism, the familiar Humean point that there are beliefs that cannot be suspended. Even if belief could be suspended, tranquility could not be achieved through equipollence. As Hume sees it, the Pyrrhonian leaves us with the problem of explaining how mental equilibrium might be achieved. As we have seen (§3), the skeptic, finding contradictions of equal weight, gave up aiming for truth, and suspended belief. In Hume’s view, this gains nothing for the skeptic, in respect to tranquility. As with Hume’s first criticism, there is no challenge to tranquility as the objective of inquiry. Hume’s conclusion is that if tranquility is to be achieved, it will have to be within a system of beliefs. Since belief aims at truth, Hume thus reinstates the lower-order objective of aiming at truth, an objective relinquished by the Pyrrhonian. One project in Book I of the Treatise is to determine whether a settled condition can be achieved within a system of beliefs, given that it cannot be achieved through an equipollence of opposing arguments where belief is suspended. Hume takes this to be a pressing issue, for he thinks that disequilibrium is a pervasive feature of our mental life. On the face of it, many stretches of the Treatise—pages 186–87, 198–210, 210–16, 219–24, 253–55, 439–48—are given over to descriptions of disequilibria in our doxastic or emotional lives. Since an unsettled condition causes uneasiness, a

51. The picture that emerges is that of a set of doxastic states that are in a state of dynamic equilibrium, but not in static equilibrium. The state is “dynamic,” owing to the presence of cycles of change; it is in “equilibrium,” in the sense that the cycles recur. (Suppose physical laws and initial conditions such that the universe alternately expands and contracts; such a universe, though far from static, would nevertheless be in dynamic equilibrium.) Hume’s insight is that there are systems of belief in “equilibrium,” dynamic equilibrium, that are not static. Hereinafter, I use ‘equilibrium’ as a shorthand for ‘static equilibrium’. 52. Hume anticipates recent objections to Pyrrhonian skepticism: cf. Burnyeat 1980, 20–23, and Annas and Barnes 1985, 170–71. Hume, however, has a more elaborate account than the contemporary criticisms of why equipollence is unpleasant. Penelhum 1983b attributes to Hume the point that suspense of judgment or doubt leads to anxiety, but Penelhum does not see that equipollence leads to disturbance because it is unsettled.

sextus, descartes, hume, and peirce

137

kind of disturbance, mental disequilibrium is itself a serious threat to the achievement of ataraxia. It is these Pyrrhonian themes and concerns, and Hume’s reaction to them, that stand in the background of Hume’s Peircean starting points.

6. Hume’s Pessimism Sextus, Descartes, and Peirce are sanguine that a settled condition can be achieved. Sextus identifies a battery of ways or modes for achieving equipollence (cf. PH I.31–186), and hence suspension of judgment and tranquility. For Descartes, unshakable belief can be achieved by attending to a clear and distinct demonstration of the existence of a nondeceiving God. For Peirce, settled belief can be achieved by employing the method of science. Hume, in the end, brings himself to a pessimistic conclusion about the possibility of achieving settled belief. Hume’s pessimism is in full view in I.iv.7, “The conclusion of this book.” This complex section has recently received a good deal of attention. I confine myself to some brief comments, from the perspective of my overall interpretation of Hume. Hume writes: “I am ready to reject all belief and reasoning, and can look upon no opinion even as more probable or likely than another” (T 268–69). Hume cannot mean that he is ready to suspend all belief; some beliefs are irresistible (31, 225; cf. 128). He means that he is ready to reject all belief as unreasonable. At pages 265–68, Hume draws on a number of considerations en route to this result. Hume develops the most important of these considerations at pages 267–68.53 His discussion draws on conclusions at I.iv.1, “Of skepticism with regard to reason,” where Hume argues that “all knowledge,” even demonstrative knowledge, “degenerates into probability” (T 180). Hume proceeds to argue that judgments of probability are subject to correction in light of the fallibility of judgment, that this correction takes the form of a reduction in the estimate of probability, and that the new judgment of probability is itself subject to correction and reduction, ad infinitum. The result of such a series of reductions would be “a total extinction of belief and evidence” (183); it would “at last reduce [the original evidence] to nothing” (184). In I.iv.7, Hume reiterates that “the understanding, when it acts alone . . . entirely subverts itself, and leaves not the lowest degree of evidence in any proposition” (267–68). In I.iv.1, the argument for the claim that the understanding subverts itself depends on the assumption that the principles by which we correct probability judgments are “apply’d to every new reflex judgment” (T 186). As Hume puts it in I.iv.7, it is only when adherence to the understanding is “steadily executed” (267) that belief is extinguished. The extinction of belief

53. I discuss another of his considerations in my 1991, §6, 1995a, §2, and 1995b, §5.

138

reflection and the stability of belief

depends on the repeated execution of corrections to probability judgments. At I.iv.1, however, Hume claims that repeated correction of probability judgments “becomes forc’d and unnatural”; “The attention is on the stretch: The posture of the mind is uneasy” (185). As Hume writes in I.iv.7, “we enter with difficulty into remote views of things” (268). The extinction of belief, then, depends on a series of corrections and reductions that does not take place. If Hume has not shown that the understanding does subvert itself, why should he be prepared to reject all belief as unreasonable? Hume’s point, I suggest, is that reflection on the result of repeated corrections to probability judgments, were we to undertake them, unsettles our current probability judgments. We assign a non-zero probability to a great many propositions. Reflection shows that repeated corrections reduce to nothing the probability that any belief is true. Since belief aims at the truth, this consideration unsettles (albeit without entirely extinguishing) our current beliefs. When one considers that a series of corrections would reduce the probability to nothing, one is less inclined to maintain one’s current probability judgments.54 This is not, however, the end of Hume’s story. At pages 267–68, intense and sophisticated reflection (about possible corrections to probability judgments) unsettles belief. This reflection, according to Hume, gives way: Most fortunately it happens, that since reason is incapable of dispelling these clouds, nature herself suffices to that purpose, and cures me of this philosophical melancholy and delirium, either by relaxing this bent of mind, or by some avocation, and lively impression of my senses, which obliterate all these chimeras. I dine, I play a game of backgammon, I converse, and am merry with my friends; and when after three or four hours amusement, I wou’d return to these speculations, they appear so cold, and strain’d, and ridiculous, that I cannot find in my heart to enter into them any farther. (T 269) Has settled belief now been achieved, in the absence of intense reflection? Unfortunately, the relatively unreflective posture when we are involved in mundane pursuits is itself unstable: At the time, therefore, that I am tir’d with amusement and company . . . , I feel my mind all collected with itself, and am naturally inclin’d to carry my view into all those subjects, about which I have met with so many disputes . . . I cannot forbear having a curiosity to be acquainted with the principles of moral good and evil, the nature and foundation of government, and the cause of those several passions and inclinations, which actuate and govern me. I am uneasy to think I approve of one object and disapprove of another; call one thing beautiful, and 54. I provide a somewhat different treatment of this material in my 1991, §6, 1995a, §2, and 1995b, §5.

sextus, descartes, hume, and peirce

139

another deform’d; decide concerning truth and falshood, reason and folly, without knowing upon what principles I proceed. (270–71) The uneasiness in this mood should lead to increased reflection, about morality, beauty, and even the principles by which we “decide concerning truth and falsehood” (the very topic that has occupied Hume in Book I of the Treatise), returning him to investigation of the understanding. It is difficult to see why he should not cycle back to the intense reflection of Treatise 268–69 and hence oscillate between the different levels. There is no level of reflection that can be sustained and hence no steady or settled belief.55 More precisely, settled belief cannot be achieved by someone who has gotten so far as the intense reflection of pages 265–68. Such a person will lapse into a more unreflective posture and in turn undertake inquiries that lead back to unsettling intense reflection. Though Hume describes the oscillations in the first person, he does not think of them as idiosyncratic. To the contrary, he wants to draw general conclusions about the role of philosophy, for himself and others (cf. T 271–74). At the same time, many are exempt from the oscillations he describes: I am sensible, . . . that there are in England, in particular, many honest gentlemen, who being always employ’d in their domestic affairs, or amusing themselves in common recreations, have carried their thoughts very little beyond those objects, which are every day expos’d to their senses. And indeed, of such as these I pretend not to make philosophers, nor do I expect them either to be associates in these researches or auditors of these discoveries. (272) Hume has provided no reason to think that such unreflective persons cannot achieve belief that is by and large settled. They will not have had their beliefs systematically unsettled by the considerations in I.iv.7, and will not be prone to the oscillations Hume describes as infecting those who are more philosophical. It is those who have engaged in intense reflection who are subject to the oscillations Hume describes. In this regard, Hume stands in marked contrast to Peirce. Peirce is in agreement with Hume that an unreflective person can achieve settled belief. A certain degree of reflection is required to unsettle belief-forming methods other than the method of science. As Peirce writes, in discussing the method of tenacity, the needed reflection is likely to occur “in some saner moment” (CP 5.378), and in discussing the method of authority, that “some individuals will be found who are raised above” the condition of not being able to “put two and two together” (5.381). Not everyone achieves the level of reflection necessary to unsettle beliefs based on tenacity or authority. Do the

55. Cf. Laird 1932, 179; Hookway 1990a, 103; and M. Williams 1991, 8–9. For a different reading, see Baier 1991, esp. 22.

140

reflection and the stability of belief

higher levels of reflection have some pride of place in the evaluation of belief-forming methods? Peirce seems to allow that he has no answer: A man may go through life, systematically keeping out of view all that might cause a change in his opinions, and if he only succeeds . . . I do not see what can be said against his doing so. It would be an egotistical impertinence to object that his procedure is irrational, for that only amounts to saying that his method of settling belief is not ours. (5.377) In this passage, Peirce is not prepared to claim that the person who clings tenaciously to belief ought to be more reflective or that he ought to adopt a different method. Peirce’s reserve here finds itself in the company of another line of thought. Peirce maintains that there is a tendency for one method to give way to another, in a sequence moving from tenacity, to authority, to the a priori method, and to the method of science. In other words, there is a tendency to greater degrees of reflection, culminating in the method of science. This is true both at the level of the individual and at the level of entire societies.56 Thus, a portion of Peirce’s analysis of belief-forming methods is given over to the method of authority’s sway, and eventual collapse, in earlier cultures and civilizations (cf. CP 5.380–82). Peirce writes in this vein: “But most of all I admire the method of tenacity for its strength, simplicity, and directness. . . . It is impossible not to envy the man who can dismiss reason, although we know how it must turn out at last” (5.386). I take the final clause to be an allusion to the ascendancy of the method of science, in the fullness of time. Though we perhaps cannot say that persons at earlier stages in the sequence ought to be more reflective, we can take note of the tendency for one method progressively to give way to another. Here we have the marked contrast with Hume. For both Peirce and Hume, an unreflective person can achieve settled belief, albeit belief that would be unsettled by greater reflection. For Peirce, however, heightened reflection leads to another method of settling belief; for Hume, it leads to oscillation. Though Hume has a constructive orientation in epistemology, he is happy to let it lead to a pessimistic conclusion. Indeed, he seems to rush into it. The foundation in I.iv.1 for the main argument leading to the pessimism in I.iv.7 seems utterly mistaken. In order to conclude that the series of corrections would reduce probability to nothing, Hume multiplies probabilities of different orders, taking the probability that a first-order probability estimate is mistaken to reduce the original estimate.57 I think that the outcome of a close examination of the considerations at I.iv.7 would be that Hume has not provided any sound reason for thinking his naturalistic project leads to pessimism. 56. I owe my awareness of this Hegelian strand in Peirce to Jeffrey Kasser. 57. See MacNabb 1951/1966, 100–102; Hacking 1978, §9, 30; and Fogelin 1985, 16–19.

sextus, descartes, hume, and peirce

141

The epistemological position I attribute to Hume—on which settled belief is the objective of inquiry—is recessive in the Treatise. There are two reasons for this. In the first place, with his own pessimism in view, Hume does not bother to elaborate the details of his favored epistemological theory. In addition, Hume’s normative epistemological project is secondary to the project of psychological explanation and the science of human nature.58 Similarly, in the case of Descartes, I identify what is but an important strand in his thinking. We have seen in §3 that to some extent Descartes backs into unshakability as the objective of inquiry, under pressure of the problem of the Cartesian circle. Thus, my claim is that the kind of position on the regulation of doxastic states that is to be found on the surface of Sextus and Peirce has a role to play, albeit a less obvious role, in Descartes and Hume.59

58. I discuss the recessive character of Hume’s position somewhat more thoroughly in 1995a, §§2–4, passim, and 1995b, §§1 and 5. 59. I am grateful to Jeffrey Kasser for helpful discussion and encouragement, and to Daniel Garber for his careful response to a version of this chapter read at the Central Division Meetings, American Philosophical Association, April 1995. I have benefited greatly from the comments of faculty colleagues—Mark Crimmins, Edwin Curley, Sally Haslanger, David Hills, James Joyce, and Eric Lormand among them—in conjunction with a presentation of a version of this essay for a colloquium at Michigan. I also thank Geoffrey Sayre-McCord and referees for Noûs for their most helpful comments and suggestions. I am especially grateful to David Velleman, who—as in other instances—has provided indispensable help in bringing this project to fruition.

142

reflection and the stability of belief

5 Integrating Hume’s Accounts of Belief and Justification

hume assigns a pivotal role to stability in understanding normativity in a variety of theoretical contexts. Hume’s distinction between the calm emotions and violent passions may be interpreted as appealing to a difference in volatility, rather than mere intensity.1 The calm emotions are stable, the violent passions volatile, and the distinction between them carries normative import; Hume identifies the prevalence of the calm emotions with “strength of mind” (T 418).2 Artificial justice is a system of conventions in the interest of stability of possession (cf. III.ii.2–3). Moral judgments arise in light of “continual fluctuation” in our situation, “in order to prevent . . . continual contradictions, and arrive at a more stable judgment of things” (581, Hume’s emphasis). I believe that stability is crucial to Hume’s theory of justified belief, as well as to his accounts of the calm emotions, artificial justice, and moral judgment. In this chapter, I argue that a stability-based interpretation of Hume’s theory of justification resolves a puzzle in regard to Part iii of Book I of the Treatise.3 In §1 of this chapter, I present a significant textual phenomenon in Part iii: Hume’s claim that association by the relation of cause and effect produces belief is often intertwined—though without his remarking on this fact—with the claim that belief based on causal inference is justified. To 1. The received interpretation of the calm-violent distinction with reference to emotional intensity is due to Árdal 1966/1989, 9, 94, 97–98, 104. My approach is suggested in MacNabb 1951/1966, 165. 2. [In this chapter, quotations of Hume are based on the Selby-Bigge and Nidditch editions of the Treatise and first Enquiry.] 3. In my 1998 [this volume, ch. 4], I place this line of interpretation in a broader historical context.

143

explain this, I offer the hypothesis that, in Hume’s view, stability plays a double role. Whether belief is justified depends on considerations of stability, and (a species of) stability is also essential to belief itself. In §2, I show that, for Hume, any belief is stable, in that it is steady or infixed. To establish that a state is a belief is thus to establish that it is stable, other things being equal. In §3, I observe that a belief, though steady in that it is infixed, might nevertheless be unstable in its influence on thought, the will, and action— owing to the presence of other beliefs with which it conflicts. I argue that the point of Hume’s distinction between justified and unjustified belief is to call attention to circumstances in which a belief, though steady, is unstable in its influence, all things considered. I then show that this perspective is useful in understanding Hume’s readiness come I.iv.7 to reject all belief, including belief based on causal inference, as unjustified. In the final section, I distinguish two versions of my stability-based interpretation. The issue is whether Hume takes the justification of a belief to be a matter of stability within the belief system of the person who holds the belief, or to depend on the belief’s stability within the belief system of a suitably reflective person.

1. A Puzzle in Regard to Part iii of Book I The titles of sections 5–10 of Treatise I.iii include “Of the nature of the idea, or belief,” “Of the causes of belief,” and “Of the influence of belief.” On any interpretation, these sections have for their subject matter the nature, causes, and effects of belief. They also include numerous passages that register Hume’s epistemic approval of beliefs based on causal inference. In I.iii.6, Hume allows that causal inference is “just”: “cause and effect . . . ’tis the only [connection or relation of objects], on which we can found a just inference from one object to another” (T 89).4 Later in I.iii.6, Hume writes of “reasonings” (93) based on the relation of cause and effect; similarly, “we are able to reason upon” (94) the relation of causation. In a footnote to I.iii.7, Hume writes: “We infer a cause immediately from its effect; and this inference is not only a true species of reasoning, but the strongest of all others” (97n.; cf. 94, 95).5 In these contexts, such terms as ‘reason’ and ‘reasoning’ function as terms of approval.6 In Part iii, the claim that causal inference is 4. This passage occurs in the course of a discussion that has been standardly interpreted as formulating a skeptical problem of induction. There is a growing consensus that, in I.iii.6, Hume is arguing that causal inference is the product of association, without intending this conclusion to have skeptical implications. Cf. Connon 1976, esp. 129–35; Beauchamp and Rosenberg 1981, ch. 2, esp. §§III–IV; Baier 1991, esp. 54–56, 65–70; Schmitt 1992, 245n.11; Garrett 1997, ch. 4; and Owen 1999, esp. 66, 74–76, 139, and ch. 6. For a dissenting view, see Winkler 1999. There are misgivings in regard to the skeptical interpretation as early as G. E. Moore 1909, 155–56, 160. 5. Cf. Beauchamp and Rosenberg 1981, 43, 63. 6. For an extended discussion of Hume’s use of this terminology, including a shift in his usage within I.iii, see my 1997, esp. 283–84, 294–96.

144

reflection and the stability of belief

justified thus arises in tandem with the claim that causal inference results in belief. Yet, Hume does not give due recognition to the fact that these claims are different. I.iii.9, “Of the effects of other relations, and other habits,” provides a striking instance of Hume’s unacknowledged intermingling of claims about belief and claims about justification. The section discusses the relations of resemblance and contiguity. Hume grants that these relations can enliven an idea and intensify an existing belief, while insisting that they do not produce belief (cf. T 107–10). In the second paragraph of the section, Hume summarizes some previous results: I have often observ’d, that, beside cause and effect, the two relations of resemblance and contiguity, are to be consider’d as associating principles of thought, and as capable of conveying the imagination from one idea to another. I have also observ’d, that when of two objects connected together by any of these relations, one is immediately present to the memory or senses, not only the mind is convey’d to its co-relative by means of the associating principle; but likewise conceives it with an additional force and vigour, by the united operation of that principle, and of the present impression. And this I have observ’d, in order to confirm by analogy, my explication of our judgements concerning cause and effect. (107) In the continuation of the paragraph, Hume raises a difficulty: But this very argument may, perhaps, be turn’d against me. . . . For it may be said, that if all the parts of that hypothesis be true, viz. that these three species of relation are deriv’d from the same principles; that their effects in inforcing and inlivening our ideas are the same; and that belief is nothing but a more forcible and vivid conception of an idea; it shou’d follow, that that action of the mind may not only be deriv’d from the relation of cause and effect, but also from those of contiguity and resemblance. But . . . we find by experience, that belief arises only from causation. (107) The third paragraph begins: “This is the objection; let us now consider its solution” (107). The second paragraph sets the problem initially under discussion in I.iii.9. To respond, Hume needs to explain why causal inference, unlike resemblance and contiguity, produces belief. This explanation occupies the third through seventh paragraphs. Paragraphs three and four discuss belief based on the senses, memory, and the relation of cause and effect. Paragraphs five and six discuss resemblance and contiguity. Paragraph seven compares the relation of cause and effect to these other relations. In the third and fourth paragraphs, Hume introduces two systems of “realities.” The first system of realities is based on the senses and memory

hume’s accounts of belief and justification

145

alone; the second system is based on custom or the relation of cause and effect and extends or supplements the first. In the discussion of the two systems, Hume seems to be saying not only that the relation of cause and effect, unlike resemblance and contiguity, produces belief, but that the relation of cause and effect produces justified belief. The evidence for this is considerable. Hume writes that the mind “dignifies” the second system, as well as the first, “with the title of realities” (T 108). He writes, more directly, that the second of the two systems is “the object . . . of the judgment” (108). He also contrasts the ideas belonging to the second system with those “which are merely the offspring of the imagination” (108). This discussion is naturally read to suggest that Hume approves epistemically of beliefs based on the relation of cause and effect. It is difficult not to be sympathetic with John Passmore’s comment on the passage: “The fact is that ‘reality’ . . . has a honorific sense.”7 To reply to the objection at hand, Hume needs to show that the relation of causation produces belief. Yet, Hume claims that causal inference produces justified belief. Hume seems to change the subject, and to do so without notice. One could go a certain distance in avoiding this reading by taking the two-systems passages in relative isolation. It might be thought that in saying that the mind dignifies the objects comprised in the second system “with the title of realities,” Hume simply means that we believe those objects exist. And it might be thought that when Hume writes of the second system as “the object . . . of the judgment,” he simply means that they are the objects of belief. There is substantial evidence against the view that Hume thus confines his claims to belief. In the first place, there is a parallel development in the fifth and sixth paragraphs, where Hume explains why resemblance does not produce belief. (Here and elsewhere, I use ‘resemblance’ to stand for the relations of resemblance and contiguity.) Hume grants that resemblance produces a state that has affinities with belief, in which we “feign” the existence of an object. In the sixth paragraph, he offers epistemic claims: “such a fiction is founded on so little reason, that nothing but pure caprice can determine the mind to form it” (T 109); “we . . . form a general rule against the reposing any assurance in those momentary glimpses of light, which arise in the imagination from a feign’d resemblance and contiguity” (110). These epistemic assessments seem curiously out of place. In the context of the objection he is considering, Hume needs to show that resemblance does not produce belief. In one compressed discussion, he also claims that the states resemblance does produce are not justified.8 In paragraphs five and six, the states that fall short of belief are unjustified, much as in paragraphs three and four, the states that constitute belief are justified.

7. Passmore 1952/1968, 101; cf. 54, 61–62. See also Kemp Smith 1941, 383–85. 8. As Fogelin notes, “Here Hume’s approach is in part descriptive and . . . part normative” (1985, 58).

146

reflection and the stability of belief

In the second place, there is recalcitrant material internal to the discussion of the two systems of realities: “’tis this latter principle [judgment], which peoples the world, and brings us acquainted with such existences, as by their removal in time and place, lie beyond the reach of the senses and memory” (T 108). “Brings us acquainted” implies success, what we call “knowledge,” not mere belief, in objects that lie beyond the reach of the senses and memory.9 This is of a piece with other formulations early in the Treatise: cause and effect is the only relation that “informs us of existences and objects, which we do not see or feel” (74), the relation that can “discover” (73) and “lead us” (89) to objects that have not been observed. In the third place, in the footnote at pages 117–18 that closes I.iii.9, Hume reinforces his contrast between beliefs based on causal inference and those “which are merely the offspring of the imagination.” Hume contrasts “probable reasonings”—attributed to “reason” (T 118n.; cf. 371n.)—with “those whimsies and prejudices, which are rejected under the opprobrius character of being the offspring of the imagination” (117n.).10 The note repeats language in the two-systems passage contrasting states that “are merely the offspring of the imagination” (108) with the products of perception, memory, and causal inference. This is evidence that Hume intends the two-systems passage to have a place in a sustained contrast between justified and unjustified belief. In the fourth place, Hume’s approval of causal inference in the two-systems passage is continuous with his favorable discussion of causal inference throughout Part iii, and well into Part iv, of Book I.11 I reviewed some examples drawn from I.iii.5–10 earlier. In I.iii.11–13, the sections on probability, Hume offers an approving discussion of “the several degrees of evidence” (T 124; cf. 153–54). In I.iii.13, Hume discusses the effect of “a long chain of connected arguments” (144). The individual arguments, where the “inference is drawn immediately . . . without any intermediate cause or effect” (144), may be “just and conclusive in each part” (144). In I.iii.15, “Rules by which to judge of causes and effects” (173), Hume writes that “it may be proper to fix

9. What “we” in more recent epistemology call “knowledge.” Hume’s “proofs” (T 124) fall within the scope of the term as I am using it. Though Hume officially reserves ‘knowledge’ for belief arising from the comparison of ideas (T 69–70, 124), he allows in the Abstract: “No matter of fact can be proved but from its cause or its effect. Nothing can be known to be the cause of another but by experience” (Abs. 654). See also the uses of cognates of ‘know’ at Treatise 103, 104, and 148. 10. [For a textual note on Treatise 371n., see “References to Hume,” this volume.] 11. For discussions—in addition to Passmore’s—that take note of Hume’s favorable attitude toward causal inference, see G. E. Moore 1909, 149–51, 154–55; Price 1940, 31–33, and 1969, Lecture 7, esp. 173–75; Kemp Smith 1941, 382–83; Connon 1976, esp. 135–37; Immerwahr 1977, esp. 58–63; Beauchamp and Rosenberg 1981, 52–55; Broughton 1983; Baier 1991, 12, and ch. 3, esp. 56–57, 65–66; Schmitt 1992, 55; my 1995a [this volume, ch. 7], §2, and 1997, esp. 283–91; and Millican 1995, 124–27. For a recent, systematic account of the matter, see Noonan 1999, 116–31.

hume’s accounts of belief and justification

147

some general rules, by which we may know” (173) when objects cause one another. He proceeds to provide eight rules (173–75), said to constitute “all the Logic I think proper to employ in my reasoning” (175; cf. 149). Hume’s language—“rules” that are “proper” for ascertaining causes and effects, and which constitute a “Logic”—reflects epistemic approval. Hume reiterates that causal inference is “just” (216) in I.iv.2. As late as I.iv.4, Hume allows that someone engaged in causal inference “reasons justly” (225). In sum, though Hume’s ostensible purpose in I.iii.9 is to explain why the relation of cause and effect produces belief, there is substantial evidence, within I.iii.9 and surrounding sections, that he takes the relation of cause and effect to produce justified belief. Other commentators have observed that Hume seems to run together a theory of belief and a theory of justified belief. Passmore writes: “what set out to be a theory of belief, in something like the ordinary sense of the word, has become, with no explicit acknowledgment of that fact, a theory of what it is ‘rational’ to believe.”12 On this view, what is initially a theory of belief—with belief resulting only from association by the relation of cause and effect—is ratcheted up to a theory of justified belief. The textual phenomena that are bothering Passmore are real enough.13 In reading Part iii, one easily gains the impression that Hume’s claim that belief arising from the relation of cause and effect is justified is somehow yoked to his claim that the relation of cause and effect produces belief. The question is how we are to explain this puzzling feature of Part iii.14 The key is to suppose that, in Hume’s view, establishing that a state is a belief is to establish that it is justified, other things being equal.15 On what grounds could Hume maintain this? It is easy to see what is required, at least schematically. We need to locate a property that is necessary for a state to constitute a belief, such that to establish that a state is a belief and thus has this property is also sufficient to establish that the belief is justified, other things being equal. In other words, there must be a property which plays a twofold role. The presence of the property must constitute a necessary condition for belief. In addition, establishing that the property is present must constitute a sufficient condition for establishing justification, other things being equal. My claim is that stability is the property that plays this dual role, one within Hume’s theory of belief, the other within Hume’s theory of justification. In light of this, I proceed in two stages. In §2, I show that, in Hume’s view, stability is a necessary condition for belief. In §3, I address the nature of Hume’s interest in stability in the context of his theory of justification. 12. Passmore 1952/1968, 62–63. 13. Price is perhaps fastening on to this puzzle when he writes that Hume “want[s] to stick to his original definition of belief as a lively . . . idea associated with a present impression by a relation of experienced constant conjunction,” but that Hume “now wants to say that it is a definition of reasonable or sensible or sane or intelligent belief” (1969, 174). 14. Passmore’s position is complex; space does not permit me to discuss it. 15. There is an ambiguity in the parsing of this key claim. I take up this matter in §4.

148

reflection and the stability of belief

2. Steadiness and Infixing in Hume’s Theory of Belief I begin with a sketch of the role of stability in Hume’s theory of belief. Tradition in Hume interpretation has it that beliefs are lively ideas. On my interpretation, beliefs are steady dispositions. It is a commonplace that Hume uses a cluster of closely related terms—‘vivacity’, ‘vividness’, ‘intensity’, and liveliness’—to characterize belief. This vivacity cluster, however, is prima facie distinct from a second cluster of terms that also has a prominent role in Hume’s discussion of belief. Ideas that constitute beliefs are ‘fast’, ‘firm’, ‘settled’, ‘solid’, and ‘steady’; such ideas possess ‘firmness’, ‘solidity’, and ‘steadiness’ (T 97, 105, 106, 108, 116, 121, 624, 625, 626, 627, 629/1.3.7.7, 631/1.3.10.10). For ease of exposition, I use ‘steadiness’ (and its cognates) to stand indifferently for all the terms in this steadiness cluster.16 Although Hume suggests that the terms in the two clusters may be used interchangeably (629/1.3.7.7), we need to disentangle them in order to attribute to Hume a coherent theory of belief. Hume contrasts steady ideas with ideas that are ‘momentary’ (110), ‘floating’ (116), and ‘loose’ (97, 106, 116, 123, 595, 624, 625; cf. 110). The terms in the steadiness cluster often refer to a kind of staying power. I suggest that steadiness plays a more fundamental role than vivacity in Hume’s theory of the nature of belief.17 I will first show that beliefs are steady and then show that they are dispositions. Hume’s conception of steadiness is closely connected to his discussions of the ways in which ideas are infixed. There is ample evidence that when an idea is infixed, the result is a belief. At Treatise 86, 109, and 225 (cf. T 99, 121), ideas are infixed with force or vigor, or enlivened, and Hume’s formula is that belief is a lively or forceful idea.18 Furthermore, Hume writes that in belief the mind “fixes and reposes itself” (624) on its conceptions, or “fixes and reposes itself in one settled conclusion and belief” (625). I suggest that infixing is a process that produces steadiness in an idea, and hence belief. In the course of the Treatise, Hume specifies particular mechanisms as ones that infix belief. In I.iii.5, Hume maintains that “belief or assent . . . always attends the memory and senses” (T 86). He writes of “custom and habit having . . . the same influence on the mind as nature, and infixing the idea with equal force and vigour” (86; cf. 225). In context, ‘nature’ refers to the senses and memory, so that custom, in addition to the senses and memory, produces steady ideas and hence infixes belief.

16. A number of other terms—‘force’, ‘strength’, and ‘vigor’—at least in some of their occurrences perhaps belong in this cluster as well. 17. For commentators who place some emphasis on the role of steadiness in Hume’s account, see Laird 1932, 88; Wolff 1960, 298–99, 309; Pears 1990, 12, 36, 44; and Baier 1991, 72–74, 80. Also see the references to MacNabb at note 24. 18. The formula, however, is not adequate. See later in this section.

hume’s accounts of belief and justification

149

Hume identifies “custom” (T 102) or “habit” (105) with “every thing . . . which proceeds from a past repetition, without any new reasoning or conclusion” (102). In the case of inference based on the relation of cause and effect, the repetition consists in the frequent observation of resembling pairs of objects (109, 225). Custom also includes the repetition of a mere idea (86, 116, 121). Repetition produces and thus explains steadiness. Hume writes that a “principle [that] has establish’d itself by a sufficient custom” also “bestows an evidence and firmness on any opinion to which it can be apply’d” (105, emphasis added). In the Appendix, Hume observes that “it must be allow’d, that the mind has a firmer hold, or more steady conception of what it takes to be matter of fact, than of fictions” (626). He adds in the next paragraph that by the association of “frequently conjoin’d” objects, “We can explain the causes of the firm conception” and that these causes “exhaust the whole subject” (626). Repetition or frequent conjunction—as well as the senses and memory—give rise to firmness or steadiness in belief.19 My claim that beliefs are steady may be strengthened on the basis of I.iii.7, “Of the nature of the idea, or belief.” Hume writes: “We may mingle, and unite, and separate, and confound, and vary our ideas in a hundred different ways; but ’till there appears some principle, which fixes one of these different situations, we have in reality no opinion” (T 96). We have not achieved opinion unless we have fixed on a particular idea; the infixing of an idea is essential to belief. Since infixing is a process that results in steadiness, steadiness must itself be essential to belief. To this point, I have been writing as if steadiness is a property of an idea. If ideas are conscious or occurrent states, however, they are obviously unsteady—they come and go, or change, abruptly. At the same time, there is substantial evidence that steadiness is crucial to Hume’s account of belief, evidence that will be reinforced as this section proceeds. What is needed is a framework that can accommodate steadiness as a property of belief. Steadiness, I suggest, is best construed not as a property of an idea, but as a property of a disposition. Hume often invokes mental dispositions or propensities that cannot plausibly be identified with occurrent states.20 Hume’s tendency to treat belief as a disposition, as well as an occurrence, is well-known.21 For example, a passage in the Appendix begins on a tack that invites taking belief to be an occurrent or conscious state: “An idea assented to feels different from a fictitious idea, that the fancy alone presents to us: And this different feeling I endeavour to explain by calling it a superior force, or vivacity, or solidity, or

19. Hume appeals to a higher-order habit (T 104–5; cf. 131, 173–74) to defend this hypothesis against the objection that it is not compatible with our forming a belief based on a single experience. 20. Cf. Wolff 1960, 293–94; Stroud 1977, 167–68; and Bricke 1980, 46–52. 21. See the references at note 24 and also Basson 1958, ch. 3, passim; Bricke 1980, 121–22 (cf. 30–31, 46–58); and Pears 1990, 50–51.

150

reflection and the stability of belief

firmness, or steadiness” (T 629/1.3.7.7). The passage continues: “This variety of terms, which may seem so unphilosophical, is intended only to express that act of the mind, which renders realities more present to us than fictions, causes them to weigh more in the thought, and gives them a superior influence on the passions and imagination” (629/1.3.7.7). Here, belief is characterized with reference to its effects. This suggests a dispositional account. There are a number of similar passages: belief “gives [ideas] more force and influence; makes them appear of greater importance; infixes them in the mind; renders them the governing principles of all our actions” (629/1.3.7.7; cf. Abs. 654 and EHU 48–52, 56–57); and the mind “is more actuated and mov’d by” beliefs than by states that fall short of belief (T 624). Such notions as “steadiness” apply more naturally to dispositions than to occurrent states. A dispositional belief is a state in virtue of which there is a tendency to display characteristic manifestations in relevant circumstances, manifestations that are, in part, effects of the disposition. These manifestations include influence on thought, the passions, and action—verbal and nonverbal behaviors, as well as internal episodes, occurrent or conscious states. We can take the claim that it is essential to belief that it is infixed, and hence steady, to apply to dispositional beliefs. This fits nicely with Hume’s explanation of steadiness, as the result of repetition or conditioning. Dispositions that are not infixed do not qualify as beliefs (cf. T 453, 629/1.3.7.7). A dispositional belief is a steady disposition to characteristic manifestations or typical effects on thought, the passions, and action.22 Such steadiness is a species of stability, the stability of a disposition that is infixed. Hume can also admit occurrent beliefs, in the sense of conscious manifestations of a dispositional belief. These occurrent manifestations include lively or vivacious ideas. Though these lively ideas are not steady, steadiness is located at the level of the underlying dispositions that they manifest. On this account, liveliness or vivacity is one of the characteristic manifestations of belief.23 It is here that terms in the vivacity cluster have their proper home. An occurrent belief, strictly speaking, is an occurrent manifestation of a dispositional belief.24 H. H. Price takes note of an apparent contradiction in Hume’s statements of his theory of belief. Hume holds that the liveliness in belief derives from a relation to a present impression, but also that an idea can be lively or 22. There is an intermediate hypothesis, that beliefs are lively ideas that are steady in their influence on thought, will, and action—that lively ideas, when they exist, have uniform effects. Given that ideas come and go, this hypothesis does not explain the steadiness in those effects. 23. One, but only one. My interpretation does not imply that a steady disposition merely to have an idea, which thus recurs, constitutes belief; there must be a disposition to the full range of effects on thought, the passions, and action characteristic of belief. 24. My discussion of Hume’s account of belief has been influenced by that of MacNabb 1951/1966, esp. 71–81. Everson 1988 adopts a related “causal” and “functional” interpretation. Price credits Hume with “a hint or suggestion” of a dispositional analysis of belief (1969, 187; cf. 165, 188). Armstrong maintains that Hume “wavers” between dispositional and occurrent accounts of belief (1973, 71).

hume’s accounts of belief and justification

151

forceful when it is not related to a present impression at all.25 Specifically, Hume holds that poetical enthusiasm does not result in belief, even though it can produce ideas that are as lively and vivacious as occurrent beliefs: [H]ow great soever the pitch may be, to which this vivacity rises, . . . ’tis still the mere phantom of belief or persuasion. . . . A poetical description may have a more sensible effect on the fancy, than an historical narration. . . . It may seem to set the object before us in more lively colours. (T 630–31/1.3.10.10) If belief is a vivacious idea, how can the lively products of poetical enthusiasm fail to count as beliefs? The answer is that the verbal and nonverbal behaviors and internal episodes that manifest dispositional belief can also arise from other sources, thereby mimicking belief. In these cases, we have pseudo-belief, what Hume calls “counterfeit belief” (123) or “the mere phantom of belief or persuasion” (630/1.3.10.10).26 Hume writes that “the least reflection dissipates the illusions of poetry” (123).27 This is a point about the steadiness of the underlying disposition. Lively ideas, no matter how intense, constitute occurrent beliefs only when they manifest a steady disposition. This dissolves Price’s apparent contradiction. My distinction between two clusters of terminology regiments distinctions required by Hume’s views.28 To provide a consistent interpretation, we must distinguish the concepts represented by the two clusters, and we must pay more than lip service to dispositional strands in Hume’s theory of belief. We must, systematically, identify beliefs with steady dispositions and distinguish between dispositional beliefs and their manifestations, occurrent and otherwise. Absent such a regimentation of Hume’s texts, there is no prospect of making sense of his discussions of poetical enthusiasm, where vivacity is held to be insufficient for belief. Also, there is no prospect of plausibly explaining how fixity and steadiness are essential to belief.29 25. Cf. Price 1969, 172–73. Price appeals to examples other than poetical enthusiasm. 26. [For a textual note on Treatise 123, see “References to Hume,” this volume.] MacNabb uses the terminology of ‘pseudo-belief’ or ‘quasi-belief’ (1951/1966, 76, 79). I extend his treatment to poetical enthusiasm, a topic he does not discuss. 27. I do not think Hume can be taken to identify belief with a persistent occurrent idea token that has a stable vivacity. A poetical enthusiasm can be sustained and thus satisfy this condition. 28. The regimentation has its limitations. A term that is typically associated with the steadiness cluster might, on a particular occasion of use, function as if it belonged to the vivacity cluster, and vice versa. 29. Stroud takes note of passages where Hume discusses belief’s “effects on the mind,” but comments: “Hume seems never to have entertained the idea that this connection between belief and the passions and the will might constitute the very difference he seeks between belief and mere conception” (1977, 74). Similarly, Pears takes note of Hume’s “remarks about the effect of belief on action” in the course of criticizing Hume for “treating [belief] . . . as a mental event or occurrence” (1990, 50). Overlooking the insights—see note 24—of Armstrong, MacNabb, and Price during the period 1951–1973, many recent commentaries do not even acknowledge dispositional elements in Hume’s account.

152

reflection and the stability of belief

Hume’s I.iii.9 discussion of the effects of the relations of resemblance and contiguity fits nicely with the interpretation on which beliefs are steady dispositions. At paragraph six, Hume begins an examination of resemblance when it operates “single” (T 109), on its own. Resemblance produces only “momentary glimpses of light” (110); “There is no manner of necessity for the mind to feign any resembling and contiguous object; and if it feigns such, there is as little necessity for it always to confine itself to the same [idea], without any difference or variation” (109). Hume sums up: “that principle being fluctuating and uncertain, ’tis impossible it can ever operate with any considerable degree of force and constancy” (109). Resemblance produces an underlying disposition that is fluctuating or unsteady and hence fails to constitute belief. Hume takes care to contrast the effects of association by the relation of resemblance with those of association by the relation of cause and effect. He has already noted in the discussion of the second system of realities that “by their force and settled order, [ideas] arising from custom and the relation of cause and effect . . . distinguish themselves from . . . other ideas” (T 108). This observation immediately precedes the explanation at paragraphs five and six of why resemblance does not produce belief. Hume returns to the contrast at paragraph seven: The relation of cause and effect has all the opposite advantages. The objects it presents are fixt and unalterable. The impressions of the memory never change in any considerable degree; and each impression draws along with it a precise idea, which takes its place in the imagination, as something solid and real, certain and invariable. (110) Again, notions related to steadiness are prominent. Whereas the effects of resemblance are “momentary,” “fluctuating,” given to “variation,” and lacking in “constancy,” the effects of the relation of cause and effect, are “fixt,” “settled,” and “invariable.”30 I previously reviewed evidence that the repetition that underpins causal inference has the advantage of giving rise to steadiness. Hume’s comparison of the relation of resemblance to that of cause and effect confirms that steadiness is essential to belief.31 Our original puzzle is to explain why the claim that causal inference results in belief and the claim that causal inference is justified frequently find themselves in close conjunction with each other in Part iii, though Hume does not explain the connection between them. In the course of arguing that resemblance does not produce belief, Hume claims that the belief-like states it does produce are not justified. Similarly, no sooner does Hume conclude that causal inference results in belief, than he concludes that causal inference is justified. He does so without calling attention to the difference in these claims. To make sense of the textual phenomena, we must locate a property 30. Baier also calls attention to these passages (1991, 72–74). 31. For additional evidence in this regard, see my 1995b, §3, and 1998, §4.

hume’s accounts of belief and justification

153

that Hume ascribes to belief and that Hume might take to be germane to justification. The only salient candidate is steadiness. Hume extensively discusses the steadiness of belief in I.iii.5–10 (T 84–123) and associated material in the Appendix. Within these stretches of the Treatise, Hume claims that infixing is essential to belief, that custom infixes belief, and that custom or repetition gives rise to steadiness. The steadiness that results from the senses, memory, or repetition is necessary for belief. This interpretation is required to make sense of Hume’s repeated statements that beliefs result from a process of infixing and that they are fast, firm, settled, solid, and steady. It is also required to explain how the results of poetical enthusiasm can exceed belief in vivacity and yet amount only to pseudo-belief. These grounds for taking Hume to hold the view that stability, in the form of steadiness, is essential to belief are independent of his theory of justification. In the second of its roles, stability is connected to justification. In paragraphs three and four of I.iii.9, the claim that causal inference is justified— that it “brings us acquainted” with objects we have not perceived and is due to the “judgment”—accompanies the claim that causal inference is due to custom or repetition. This discussion is followed by the explanation at paragraphs five and six of why resemblance does not produce belief, an explanation that stresses the unsteady character of dispositions arising from these relations and hence confirms the role Hume assigns to steadiness. In the course of this discussion, Hume writes that there is “little reason” (T 109) to feign objects based on these relations and that we form a “general rule” (110) against doing so. When Hume writes in paragraph seven of I.iii.9 that, in comparison to the relation of resemblance, “The relation of cause and effect has all the opposite advantages,” the “advantages” he has in view are twofold, reflecting stability’s interrelated roles in his theories of belief and justification. This account serves to explain the intimate connection between claims about belief and claims about justification in Part iii. The explanation is that to establish that a state is a belief and hence stable in the sense that it is steady or infixed is sufficient to establish that the belief is justified, other things being equal. My claim that whether a belief is justified depends on considerations of stability thus integrates Hume’s theory of justification with his theory of belief. It remains, however, to consider in more detail the role of stability in Hume’s theory of justification and its relationship to the steadiness of belief.

3. The Importance of a Distinction between Justification, Other Things Being Equal, and Justification, All Things Considered Why does stability, other things being equal, matter to Hume? In the second and third paragraphs of I.iii.10, Hume explains why steadiness is important:

154

reflection and the stability of belief

Nature . . . seems to have carefully avoided the inconveniences of two extremes. Did impressions alone influence the will, we should every moment of our lives be subject to the greatest calamities; because, tho’ we foresaw their approach, we should not be provided by nature with any principle of action, which might impel us to avoid them. On the other hand, did every idea influence our actions, our condition would not be much mended. For such is the unsteadiness and activity of thought, that the images of every thing, especially of goods and evils, are always wandering in the mind; and were it mov’d by every idle conception of this kind, it would never enjoy a moment’s peace and tranquillity. Nature has, therefore, chosen a medium, and has neither bestow’d on every idea of good and evil the power of actuating the will, nor yet has entirely excluded them from this influence. Tho’ an idle fiction has no efficacy, yet we find by experience, that the ideas of those objects, which we believe either are or will be existent, produce in a lesser degree the same effect with those impressions, which are immediately present to the senses and perception. The effect, then, of belief is to raise up a simple idea to an equality with our impressions, and bestow on it a like influence on the passions. (T 118–19) Were we moved to action only by sense impressions and memories, we could not make inferences to future events and thus could not make plans to avoid prospective pain or to enjoy prospective pleasure. Were we moved to action by every idea of pain or pleasure, we would not pursue a coherent plan of action over time, since our mental activity is unsteady, our ideas wander.32 Nature therefore provides a medium between these two extremes, so that some—but not all—ideas influence the will and action. These ideas are beliefs, nature’s provision for a steady influence on the will, and hence on action. Hume’s conception of the natural function of belief helps to explain the importance he attaches to the distinction between establishing that belief is justified, other things being equal, and establishing that belief is justified, all things considered. As we shall see later, this distinction is critical to understanding the relationship between Hume’s favorable attitude toward beliefs based on causal inference in Part iii and his negative or pessimistic assessment of these beliefs in I.iv.7. Any belief is steady in its influence on thought, the passions, and action in the sense that it is infixed or steady in virtue of the mechanism that produces it. It does not follow that the belief is steady in its influence, all things considered.A belief might fail to be steady in its influence owing to the presence of beliefs with which it conflicts, beliefs which reduce the likelihood of the occurrence of its characteristic manifestations or its typical effects—which reduce its influence on the will and action. 32. Winters also draws on Treatise 118–19 to make this point (1981, 640–41).

hume’s accounts of belief and justification

155

In Hume’s view, the point of a distinction between establishing that belief is justified, other things being equal, and establishing that belief is justified, all things considered, is to call attention to the kinds of circumstances in which belief, a steady disposition, might nevertheless be unsteady in its effects. The “other things being equal” qualification is thus cashed out substantively, with reference to conditions in which states are infixed but nevertheless unsteady in their influence due to the operation of other mechanisms. Hume’s concern is with circumstances that undermine the natural function of belief. Hereafter, I use the term ‘stable’ as a shorthand for “steady in its influence on thought, passions, and action.” There are two principal ways in which a belief, though steady in that it is infixed, could be unstable, unsteady in its influence. First, a belief might be unstable owing to the presence of a contradictory belief. In I.iv.2, “Of skepticism with regard to the senses,” Hume discusses two related sets of contradictory beliefs. Hume characterizes “contradiction” as an “opposition” or “combat” (T 205–6); “contradiction” involves “struggle and opposition” between “two enemies” (215), in an effort to “destroy” (215) one another.33 The language of psychological conflict is not merely metaphorical or figurative; the conflicts have identifiable effects. Hume writes: “not being able to reconcile these two enemies, we endeavour to set ourselves at ease as much as possible, by successively granting to each whatever it demands” (215). In holding contradictory beliefs, we are conflicted at the level of our underlying dispositions.34 We might imagine that contradictory beliefs cancel each other out, or that they are locked in an ongoing standoff for influence. Hume maintains, to the contrary, that contradictory beliefs typically alternate in their influence, with one combatant, then the other, temporarily gaining the upper hand.35 These dispositions are infixed and steady, but—owing to the presence of the opposing belief—neither is stable. There is a second kind of case in which a belief might be unstable owing to the presence of beliefs with which it conflicts. The influence of a belief might be undermined by the presence of second-order beliefs that one takes to call into question whether a first-order belief is true. An example would be a belief system that includes contradictory beliefs and also the second-order belief that one believes this contradiction. To take another example, suppose one believes that p and also holds the second-order belief that the belief that p results from a belief-forming mechanism that is

33. For additional detail, see my 1991, esp. 248–53. 34. In my 2001a [this volume, ch. 6], I argue that Hume seeks to ground an account of conceptual confusion or “quasi-content” in the propensities that give rise to such conflicts. 35. Baier claims that for Hume contradictions and “contrariety in beliefs is . . . essentially a matter of threat of mutual destructiveness” (1982, 645; cf. 644–47). (There are traces of this position in her 1991, 6–20.) Hume defines “contrariety” as a relation of ideas which is “discoverable at first sight” (T 70) and “susceptible of certainty and demonstration” (463). At the same time, Baier is on the right track; Hume’s real interest is in “a more dynamic conception of destructive psychological force” (1982, 645).

156

reflection and the stability of belief

unreliable, leading to false beliefs more often than not. In these and related instances, one takes one’s second-order beliefs to indicate either that some of one’s beliefs are not true, or not likely to be true. Hume, I suggest, holds that reflection on such second-order beliefs unsettles the first-order beliefs that are called into question, so that one is less inclined to maintain them. The first-order beliefs remain sufficiently infixed to constitute belief, but are like a steady performer who becomes shaken or rattled, or suffers a loss of confidence. To say that the second-order belief “unsettles” the firstorder belief is to say that it unsteadies it—in the sense that its steadiness is somewhat reduced.36 The unsettled belief, though infixed and steady, is less likely to produce its typical effects, less likely to influence the will and action. Part iv of Book I contains a number of case studies of reflections that call into question the truth of one’s beliefs. Here I have in view considerations Hume raises in I.iv.1 and I.iv.4 and elaborates in I.iv.7. Hume writes in the final paragraph of I.iv.4, “Of the modern philosophy”: “there is a direct and total opposition betwixt our reason and our senses; or more properly speaking, betwixt those conclusions we form from cause and effect, and those that persuade us of the continu’d and independent existence of body” (T 231). Hume claims that the senses lead us to believe that matter has a continued and independent existence and that causal reasoning shows that this belief is false. Hume appeals to this contradiction between the senses and causal reasoning in I.iv.7: ’Tis this principle [the imagination, or the vivacity of ideas], which makes us reason from causes and effects; and ’tis the same principle, which convinces us of the continu’d existence of external objects, when absent from the senses. But tho’ these two operations be equally natural and necessary in the human mind, yet in some circumstances they are 1directly contrary. (266) Hume’s footnote is to I.iv.4. The passage continues: “How then shall we adjust those principles together? Which of them shall we prefer?” (266). What Hume describes here, in part, is the cycle of alternating influence that can occur in the presence of contradictory beliefs. There is, moreover, an additional source of instability once a contradiction is recognized. Reflecting on the contradiction, we have a need to “adjust those principles together”; we are disposed to adjust our principles or beliefs. In these circumstances, the first-order beliefs are themselves unsettled. Though these beliefs are steady in that they are infixed, when one considers the contradiction, one is less inclined to maintain them. This unsettles the dispositions in question. 36. Here I prefer ‘unsettle’ to ‘unsteady’, because the latter admits an alternative reading, such that to “unsteady” a belief is to render it not steady, sufficiently unsteady that it no longer constitutes belief.

hume’s accounts of belief and justification

157

In I.iv.1, “Of skepticism with regard to reason,” Hume also introduces a consideration that calls into question the truth of one’s beliefs. Indeed, he claims to identify a systemic source of instability. Hume argues that “all knowledge,” even demonstrative knowledge, “degenerates into probability” (T 180). “Probability” is “that evidence, which we employ in common life” (181), evidence based on causal inference. Hume argues that judgments of probability are subject to correction in light of the fallibility of judgment, that this correction takes the form of a reduction in the estimate of probability, and that the new judgment of probability is itself subject to correction and reduction, ad infinitum. The result of such a series of reductions, he claims, would be “a total extinction of belief and evidence” (183); it would “at last reduce [the original evidence] to nothing” (184).37 Hume takes up this material in I.iv.7. At page 267, he frames “a very dangerous dilemma.” We can rely on the understanding—demonstrative and causal inference—alone, subjecting probability judgments to repeated corrections. Or we can rely on the understanding together with “seemingly trivial” (T 268) properties. Hume writes of the first alternative: “I have already shewn,1 that the understanding, when it acts alone, . . . entirely subverts itself, and leaves not the lowest degree of evidence in any proposition” (267). Hume’s footnote is to I.iv.1.38 If, however, we admit the trivial propensities, we embrace “manifest absurdities” (268). Recognizing this dilemma—much as in the case of noticing a contradiction—we have a need to adjust our faculties and beliefs: “What party, then, shall we choose among these difficulties?” (268). And much as Hume cannot find a way to revise his beliefs when he attends to the contradiction between the senses and causal inference, he writes of the present dilemma: “For my part, I know not what ought to be done in the present case” (268). As it stands, the dilemma is artificial. It is not open to us to adhere to the understanding alone. The extinction of belief depends on the assumption

37. Owen contends that Hume’s “concern is not about justification, but about truth” (1999, 189). On most any account of justification, there is some connection between justification and truth. This observation also bears on Garrett’s position that the line of development at I.iv.I and pages 267–68 of I.iv.7 unfolds entirely within cognitive psychology, rather than normative epistemology (cf. 1997, 222–27). The corrections are for the likelihood of “error” (T 180, 182) in light of “our fallible . . . faculties” (180). For related criticism of Garrett, see Fogelin 1998, 168. 38. There is the suggestion in I.iv.1 that the argument that probability reduces to nothing is directed at non-Humean conceptions of belief: “If belief, therefore, were a simple act of the thought, without any peculiar manner of conception, or the addition of a force and vivacity, it must infallibly destroy itself” (T 184). This suggestion is canceled, twice over. In the first place, Hume writes in the next paragraph: “But here, perhaps, it may be demanded, how it happens, even upon my hypothesis [that belief is more properly an act of the sensitive, than of the cogitative part of our natures], that these arguments above-explain’d produce not a total suspense of judgment” (184). In the second place, at Treatise 267 Hume appeals to I.iv.1 to support a general result about the subversion of belief. Cf. Lynch 1996, §II, esp. 100–101, and Owen 1999, 198.

158

reflection and the stability of belief

that the principles by which we correct probability judgments are “apply’d to every new reflex judgment” (T 184). It is only when adherence to the understanding is “steadily executed” (267) that belief is extinguished. Hume has argued that repeated correction of probability judgments “becomes forc’d and unnatural” (185); “The attention is on the stretch: The posture of the mind is uneasy” (185) and “we enter with difficulty into remote views of things” (268). Though we might “take a resolution” to adhere to the understanding alone, this is not a resolution we could keep. The extinction of belief depends on a series of corrections and reductions that does not take place. Hume’s point, I suggest, is that the considerations in I.iv.1 generate the second-order belief that were we to subject judgments of probability to repeated corrections, the probability of the original judgment would reduce to nothing. This second-order belief calls into question the truth of probability judgments and thereby unsettles them. When one considers that a series of corrections would reduce to nothing the probability that first-order beliefs are true, one is less inclined to maintain one’s current probability judgments. As we have seen, the second-order belief that we hold contradictory beliefs, based on the senses and causal inference, about the existence of matter also unsettles belief. In sum, I.iv.7 may be read as calling attention to the unsettling effects of some second-order beliefs which call into question the truth of other beliefs one holds. Hume takes the instabilities that emerge in I.iv.7 to bear negatively on the justificatory status of the beliefs that are unsettled. He discusses the unsettling effects of considerations introduced earlier in Part iv at pages 265–68 of I.iv.7. At page 268, Hume writes: “I am ready to reject all belief and reasoning.” Hume cannot mean that he is ready to suspend all belief; some beliefs are irresistible (T 31, 225; cf. 128). He means that he is ready to reject all belief as unjustified. Similarly, he writes that he “can look upon no opinion even as more probable or likely than another” (268–69). Thus we find in Hume the judgment that no belief, not even belief based on causal inference, is justified. The considerations that lead Hume to this result are ones that call into question the truth of his beliefs, thereby unsettling them. This reduces the steadiness in influence these beliefs would otherwise have, in virtue of being infixed, and thus undermines the natural function of belief, at least for the reflective person. This is, at least, one reading of I.iv.7, Hume’s difficult “Conclusion of this book.” A stability-based interpretation provides a fruitful account of the relationship between Parts iii and iv of Book I. I have argued that the interpretation explains why the claims that causal inference results in belief and that causal inference is justified are entwined in I.iii.9 and neighboring sections. Whereas Hume registers epistemic approval of causal inference throughout Part iii, in I.iv.7 he is “ready to reject all belief and reasoning.” Whereas Hume provides an extended account of “the several degrees of evidence” (T 124; cf. 153–54) in I.iii.11–13, in I.iv.7 he “can look upon no opinion even as more probable or likely than another.” Hume’s destructive conclusions in I.iv.7 represent a complete reversal in his attitude toward

hume’s accounts of belief and justification

159

causal inference in comparison to Part iii of Book I.39 To establish—as Hume does in Part iii—that states are beliefs, and hence steady in their influence in that they are infixed, is to establish that they are justified, other things being equal. To establish—as Hume does in Part iv—that beliefs are unstable, is to establish that they are not justified, all things considered. Hume’s readiness to reject all belief as unjustified emerges in light of his claim that beliefs based on causal inference are unstable. My interpretation thus accommodates the destructive conclusions in Part iv, as well as explaining the favorable results of Part iii.

4. Two Versions of a Stability-based Interpretation There is, however, an ambiguity in my account of the relationship between Part iii and Part iv. I have said that the instabilities that emerge in I.iv.7 “bear negatively” on the justificatory status of beliefs based on causal inference. Would Hume want to say that belief based on causal inference is never justified because it is susceptible to the instabilities he identifies in Part iv? Or might beliefs based on causal inference be justified, even though susceptible to these instabilities, provided they are not in fact infected by them? These questions are closely related to an ambiguity in my distinction between establishing that a belief is justified, other things being equal, and establishing that a belief is justified, all things considered. Let us say that a person who searches for contradictions among beliefs, examines the reliability of belief-forming mechanisms in producing true beliefs, considers the results of methodical application of the cognitive faculties, and so forth, is (fully) reflective. There are degrees of reflectiveness. For present purposes, it will suffice to group together persons who are not fully reflective as unreflective. With this classification in hand, we can explain the ambiguity I have in mind. On one interpretation, a belief is justified, all things considered, if it would be steady in its influence even within the belief system of a reflective person, that is, when fully examined in the ways I have characterized. To say that a belief is justified, other things being equal, is to say that for all that has been shown the belief would be steady in its influence for a reflective person. On this interpretation, to establish that a state is a belief is to establish-otherthings-being-equal, to establish provisionally, that it is justified. Any gap between establishing justification “other things being equal” and establishing justification “all things considered” is a gap in our knowledge of what a reflective person’s belief system would be like.

39. Cf. Passmore 1952/1968, 54–64, 99–101; Immerwahr 1977; Broughton 1983, esp. §§3 and 5; Schmitt 1992, ch. III; Millican 1995, 134; and my 1995a, §2. For a different account of developments in Part iv, see Baier 1991, 4–8, 12, 284–85. I discuss Baier’s position in my 1994.

160

reflection and the stability of belief

On the second interpretation, a belief is justified, all things considered, if it is steady in its influence within the belief system of the person who holds the belief, given the degree to which the person who holds the belief is reflective. To say that a belief is justified, other things being equal, is to say that it is steady, infixed by the senses, memory, or repetition. Such a belief might nevertheless be unsteady in its influence, all things considered, within the belief system of the person who holds the belief, owing to the presence of other beliefs with which it conflicts. On this interpretation, to establish that a state is a belief is to establish that it is justified-other-things-being-equal. Any gap between establishing justification “other things being equal” and establishing justification “all things considered” is a gap between the steadiness that accrues to any belief and the steadiness in its influence of a belief that does not conflict with other beliefs one holds. In light of Hume’s conclusions in I.iv.7, the two versions of a stabilitybased interpretation diverge in their epistemic assessments of the unreflective person. On the first version, the stable beliefs of an unreflective person are unjustified because they would be unsteady in their influence within the belief system of a reflective person. On the second version, the stable beliefs of the unreflective person are justified because, given the degree to which the person is reflective, they are steady in their influence; it is the reflective person who is in the grip of instability. Both versions assign a crucial role to considerations of stability; they differ in what that role is. On the second version, justification depends on whether a belief is stable within the actual belief system of the person who holds the belief. This version imposes a requirement of actual or de facto stability. On the first version, justification depends on whether a belief would be stable within the belief system of a reflective person. This requirement of stability under reflection is more demanding or stringent than that of stability in one’s actual belief system.40 It might be helpful to consider the character of the second version of Hume’s theory of justification, where the requirement for justification is stability within a person’s actual belief system. According to this version, to say that a belief is justified, other things being equal, is to say that it is stable, steady in its influence, simply in virtue of being infixed, simply in virtue of being a belief. This amounts to saying that it is an intrinsic property of beliefs, a property they automatically have in virtue of being beliefs, that they are justified. This thesis is characteristic of a “negative coherence

40. Reflexive approval interpretations take Hume to identify a normative mental operation or faculty with one that bears self-scrutiny, or that delivers a positive judgment when it reflects upon itself. See Korsgaard 1989 and 1996, 60–65; Baier 1991, 15, 55, 58, 93, 97, 99–100, 196–97, 277, 282, 284–85. These interpretations identify normativity with the capacity to survive reflective scrutiny (cf. Korsgaard 1989, pt. 1, and Baier 1991, 99). The first version of a stability-based interpretation has affinities with these interpretations. The second version, however, allows that an unreflective person holds beliefs that are justified though they would not survive self-scrutiny or reflection; it is therefore not an instance of a reflexive approval view.

hume’s accounts of belief and justification

161

theory” of justification.41 In a “positive” coherence theory, beliefs are justified only if they cohere with other beliefs. Coherence is required to make a positive contribution to justification, or else there would be no justification at all. In a “negative” coherence theory, one is automatically justified in any belief one holds. This justification is prima facie or defeasible. Considerations of coherence function negatively, in that incoherence can undermine or defeat this automatic justification. The requirement of de facto stability yields a version of a negative coherence theory, with Hume identifying “incoherence” with sets of beliefs that conflict in the sense of reducing the influence of beliefs that are otherwise steady.42 A standing difficulty for negative coherence theories is to explain how one’s beliefs can be automatically justified, how justification, even prima facie or defeasible justification, can be an intrinsic property of belief. A negative coherence theory seems committed to some kind of “self-justification” for every belief, and at the same time deprived of such foundationalist resources as incorrigibility that might be thought to provide a source of selfjustification. Hume has something to offer here: beliefs are automatically or intrinsically justified because they are a steady influence on the will and action—absent other beliefs that reduce that influence—simply in virtue of being beliefs. If justification, all things considered, is a matter of steady influence, then beliefs are automatically justified, other things being equal. The two versions of the stability-based theory lead to different understandings of the relationship between Parts iii and iv of Book I of the Treatise. Let us begin with the first and more demanding version, which requires stability under reflection. On this interpretation, when Hume registers epistemic approval of beliefs based on causal inference in Part iii, he is claiming that, so far as he has examined causal inference, he has not uncovered any considerations to suggest that beliefs based on causal influence would be unstable if subjected to greater reflection. The only considerations in play in Hume’s discussion of the nature and causes of belief are ones that relate to custom or habit, the mechanism that produces and infixes the dispositions that arise from causal inference. Indeed, as of I.iii.10, Hume has not fully scrutinized custom or repetition itself; in I.iii.12, he introduces a distinction between “imperfect” (T 131) and “perfect” (134, 135) habits. What is more, as of the close of Part iii, Hume has not considered either methodical application of causal inference, or the belief in body, or the relationship between causal inference and the senses, and so forth. 41. I have in view John Pollock’s distinction between “negative” and “positive” coherence theories (1979, esp. §IV, 105–11, and 1986, esp. 72–73, 83–87). 42. Elsewhere, I develop an interpretation where a belief’s justification depends, not on its own stability, but rather on the tendency of the mechanism that causes it to produce stability in belief (2002, ch. III). Elaborated in this way, Hume’s position does not imply that beliefs are justified, other things being equal, simply in virtue of being beliefs. Rather, it implies that those beliefs that result from psychological mechanisms that tend to produce beliefs are automatically justified. I am indebted to Frederick Schmitt for helping me distinguish the implications of the two developments of my position.

162

reflection and the stability of belief

Hume defers such questions to Part iv. In I.iv.7, he appeals to the instabilities located in I.iv.1 and I.iv.4 en route to his destructive conclusions about justified belief. Thus, an interpretation on which Hume holds the first and more demanding theory provides a natural reading of the recurrent claim in Part iii that causal inference is justified. This is a provisional judgment; for all Hume has shown at that stage, beliefs based on causal inference are stable—so far, so good. Once the considerations advanced in I.iv.7 are in view, this provisional judgment is withdrawn in favor of the “all things considered” judgment that belief based on causal inference is unjustified.43 Such belief is unjustified because it would be unstable for a reflective person. This is a reading of the relationship between Parts iii and iv from the perspective of an interpretation that ascribes to Hume the requirement of stability under reflection. The interpretation that attributes to Hume the less demanding theory can also take on board the developments in Part iv that I have reviewed. If Hume’s requirement is stability in one’s actual belief system, we can think of his discussion of the two systems of realities—based on the senses, memory, and custom—as exhibiting a specimen belief system someone might hold. The beliefs it contains are automatically justified, other things being equal, simply in virtue of being beliefs. Given that the subject does not engage in the sorts of reflections characteristic of Part iv, there is no tendency to incoherence or instability that would undermine the prima facie or defeasible justification within this belief system, so that the beliefs it contains are also justified, all things considered. This is very different from the upshot of attributing to Hume the more demanding theory. If justification requires stability under reflection, beliefs based on the senses, memory, and causal inference are—as of Part iii—merely justified for all that has been shown. Since instabilities would emerge upon reflection, the beliefs are not justified, all things considered. On the less demanding reading, if there is no instability in the specimen belief system, all the considerations relevant to that person’s belief system have been taken into account; the other things being equal condition is satisfied, and the person has justified belief, all things considered. Hume’s position, I have argued, is that to establish that a state is a belief, and hence steady, is sufficient to establish that it is justified, other things being equal. There are two versions of this interpretive thesis. Both afford a natural resolution of the original puzzle—to explain why the claim that causal inference is justified arises in tandem with the claim that causal inference results in belief. On the more demanding reading, to establish that the products of causal inference are beliefs, and hence steady in that they are infixed, is to establish that they would be steady in a reflective person’s belief 43. There are further developments in I.iv.7, beyond pages 268–69. I believe the main outcome is to exhibit additional instability, in that once someone has entered into the reflections of I.iv.7, he will oscillate between periods in which his beliefs are unsettled by reflection and those in which they are not.

hume’s accounts of belief and justification

163

system, for all that has been shown; the claim in Part iii that causal inference is justified is provisional. On the less demanding reading, to establish that the products of causal inference are beliefs, and hence steady in that they are infixed, is to establish that these beliefs are justified, other things being equal.44 I believe one must adopt one or the other of these interpretations— interpretations that link justification to stability—in order to explain the textual phenomena in Part iii.45

44. I defer consideration of evidence that Hume favors one of the two theories over the other. See my 2002, esp. ch. III. 45. I have been fortunate to have colleagues who have generously provided invaluable comments on my work: David Hills over the course of nearly my entire career beginning in graduate school, David Velleman for two decades at Michigan, and in more recent years Jim Joyce. This chapter owes much to them. I also thank participants in the 1999 Hume Society Conference and an anonymous referee for helpful comments. I completed this essay while enjoying a year as a Fellow at the Center for the Study of Modern Philosophy, University of Massachusetts, Amherst. I am grateful to the Center and the University of Massachusetts Department of Philosophy for gracious support and to the University of Michigan for sabbatical leave.

164

reflection and the stability of belief

6 Hume’s Explanations of Meaningless Beliefs

in this chapter I raise some puzzles in regard to Hume’s reliance on empiricism about meaning. Most fundamentally, Hume’s arguments to meaninglessness are often too strong for his purposes. They are accompanied by psychological explanations of why we believe in the existence of entities that fall under the meaningless expressions, explanations which presuppose the meaningfulness of the relevant terms. Drawing on his treatment of the notion of substratum and other examples, I propose a solution based on resources in Hume’s texts. I also apply the solution to suggest a reconstruction of Hume’s treatment of necessary connection.

1. Some Puzzles about Hume’s Meaning Empiricism The puzzles I have in view are best introduced against the background of two theses from Locke’s An Essay concerning Human Understanding: first, the meaning of a word, term, or expression in its “primary or immediate Signification” (Essay III.ii.2) is the idea for which it stands; and second, all ideas are ultimately derived from simple ideas, those received passively in experience.1 Taken together, these theses constitute Locke’s “empiricism” about meaning. It is an upshot of Locke’s theory that an expression is meaningless if it does not stand for an idea derived from experience (III.ii.2, 7). 1. [In this chapter, quotations of Hume are from the Selby-Bigge and Nidditch editions of the Treatise and first Enquiry. Also, in-text notes have been converted to footnotes, so that note numbers do not correspond to the original article. The section titles have been added.]

165

One tendency in Locke is to employ meaning-empiricism destructively against the notion of substratum. A material substratum is supposed to be a subject or substance in which the qualities of a material object—extension, figure, and so forth—inhere or subsist. In the first three editions of the Essay, Locke is emphatic that we have “no Idea” (I.iv.18) of substratum. It is difficult not to read a number of other sections (II.xiii.18–19, II.xxiii.2, 15) as registering at least ambivalence with respect to the meaningfulness of a notion of substratum.2 On the face of it, Hume is committed to Locke’s empiricism about meaning, as the earliest sections of A Treatise of Human Nature show. At SelbyBigge page 2, Hume introduces an account of the relation between simple and complex ideas. At pages 2–3, he claims that simple ideas copy experience. At pages 15–16, he applies his Lockean theory of meaning destructively to the notion of substance, concluding that “We have . . . no idea of substance, distinct from that of a collection of particular qualities, nor have we any other meaning when we either talk or reason concerning it” (T 16). Empiricism about meaning is also in place in An Enquiry concerning Human Understanding. In Section II, Hume introduces a distinction between simple and complex ideas, claims that simple ideas copy experience, and adds about meaning, “When we entertain, therefore, any suspicion that a philosophical term is employed without any meaning or idea (as is but too frequent), we need but enquire, from what impression is that supposed idea derived?” (EHU 22). Though Hume does not mention specific applications, he does draw a general moral about meaning, and does so, presumably for emphasis, in the final paragraph of the section. There are, however, a number of puzzles about Hume’s deployment of his theory of meaning. As a general observation, his destructive arguments to meaninglessness are cursory in character: they tend to occupy single paragraphs, often filled with rhetorical challenges, but lacking in sustained argument.3 There are a number of other manifestations of his taking a short way with these arguments. For example, according to Locke, complex ideas are derived from simple ideas by combining or compounding, comparing or relating, and separating (Essay II.xii). By contrast, Hume consistently treats complex ideas as formed only by compounding parts (T 2; EHU 18–20, 47). Hume’s version of the simple-complex distinction seems to be a retrogression from Locke’s account, allowing fewer resources in the form of permissible psychological operations for deriving complex ideas. What is more, in his destructive arguments to meaninglessness, Hume sometimes writes as if he is bent on finding a single perception from which a target idea is derived.4 (By a target idea, I mean an idea whose existence is in question.) En route to the claim that we have no idea of the self or soul, 2. Cf. Bennett 1987, esp. §IV. 3. Cf. Craig 1987, 91, 123–28. 4. See Aune 1991, 62.

166

reflection and the stability of belief

he observes, “It must be some one impression, that gives rise to every real idea. But self or person is not any one impression, but that to which our several impressions and ideas are suppos’d to have a reference” (T 251). To the extent that Hume searches for a single impression that is the source of the target idea, he seems to overlook the possibility that the idea of the soul is complex. It is the simple constituents of complex ideas, not the complex ideas themselves, that need to represent simple impressions exactly.5 Why is Hume content to advance destructive applications of his empiricism about meaning which are so perfunctory? Barry Stroud has suggested that Hume “adopts [the theory of ideas] without criticism from his predecessors” and that this accounts for his “quick, not very careful or thorough, exposition of the theory.”6 In this spirit, it might be suggested that Hume can make do with a truncated version and streamlined applications specifically of the Lockean theory of meaning. Locke’s empiricism about meaning would be familiar to readers. Hume could rely on Locke’s work without providing detail that would otherwise be necessary. There is, however, a nastier problem. At pages 15–16 of the Treatise, Hume declares ‘substratum’ meaningless. In I.iv.3, he sets out to explain why the ancient philosophers believe in the existence of material substrata. These sections work at cross-purposes. How can Hume consistently set out to explain the psychological causes of a belief that is without meaning or content in the first place? As Robert Fogelin writes, “What is the content of the false philosopher’s belief in substance? Hume’s answer seems to be that it is contentless, but then what does the belief amount to?”7 Granted, Hume might try to explain why the ancient philosophers believe that their belief is meaningful. Fogelin observes that Hume does try to explain this at Treatise 224: “it naturally happens, that after the frequent use of terms, which are wholly insignificant and unintelligible, we fancy them to be on the same footing with [significant and intelligible terms].”8 That is fine as far as it goes, but Hume also offers an explanation of why the ancient philosophers believe that substrata exist; if the term ‘substratum’ is “wholly

5. Craig notes that “Hume spends no time on an issue that must surely be absolutely vital to his argument: whether the idea of the self might not be a complex idea” (1987, 119). The puzzle is not simply that Hume does not consider whether the idea is simple or complex, but that he seems to assume it is simple. If so, Craig’s explanation (120), that Hume’s real concern is with the evidence for the belief, and hence with the existence of an actual impression (whether simple or complex) of the self, is not fully responsive to the problem. 6. Stroud 1977, 17. 7. Fogelin 1985, 11–12; cf. 7. For the puzzle in the related context of immaterial substance, cf. Craig 1987, 114. Stroud writes in regard to immaterial substance: “The fiction of a substance is also unintelligible, according to Hume, since it requires us to have an idea of something of which no idea can be formed. . . . Furthermore, we do not need the notion of substance in order to explain how we come to attribute identity to things” (1977, 120–21). But Hume does need some notion of substance for the purpose of his explanation of how we come to believe in the existence of a soul (T 253–55). 8. Fogelin 1985, 12. See also his 1993, 111.

hume’s explanations of meaningless beliefs

167

insignificant and unintelligible,” it is not possible to identify the belief Hume seeks to explain. This nastier difficulty cannot be dismissed on the ground that Hume is relying on Locke’s results. It involves an apparent inconsistency in Hume’s system: on the one hand, he declares the term ‘substratum’ meaningless; on the other, he provides a psychological explanation of why the ancient philosophers believe that substrata exist. The latter project is important to him, so much so that he supplements his explanation of the belief in substrata with a separate explanation of the belief in their simplicity (T 221). We also cannot dismiss the difficulty on the ground that it merely infects Hume’s discussion of substratum. Hume offers explanations of the causes of a number of beliefs formulated with respect to expressions which he has declared meaningless: beliefs in material substrata, immaterial substrata or souls, external existence, and necessary connection, among others. What I have called “the nastier problem” is far from isolated. This observation generates a variant of the main difficulty. All contentless concepts are the same, just as there is only one null set. Yet Hume provides different psychological explanations of the beliefs in the existence of material substrata, souls, external existence, and necessary connection. The different explanations can be appropriate only if the beliefs somehow differ in content, but they do not differ in content if the key concepts are meaningless. Perhaps strict meaningfulness requires ideas derived from experience in accordance with Lockean empiricism about meaning. The relevant concepts must nevertheless possess some surrogate for meaning, or some content-like features—quasi-content—if Hume’s psychological explanations of these beliefs are themselves to make sense.

2. Belief in Substrata and a Pattern of Psychological Explanation I believe we can extract an account of quasi-content from Hume’s explanation of the belief in material substrata. The explanation takes the form of a four-stage psychological reaction. Hume considers an uninterrupted succession of changing, sensible qualities. Descartes’ piece of wax slowly changes its shape, hardness, color, and other qualities as it melts. Hume notes that we can consider the succession of qualities in either of two ways. One way is to trace the sensible qualities gradually through time. Hume holds that there is a concept of identity that requires an object which has an uninterrupted and an unchanging existence (T 200–201, 219–20, 253, 255).9 Observing the changes gradually, we are tricked by an imaginative tendency or propensity

9. He thinks this is the notion of identity in its strict sense (T 200–201, 219–20, 253, 255). Here I bracket issues about how one acquires the idea of identity. See, for example, Fogelin 1985, 70–73.

168

reflection and the stability of belief

to ascribe identity to related objects.10 The successive qualities of the object are related—uninterrupted and closely resembling. This relation is similar to that of being uninterrupted and perfectly resembling or unchanging. In light of this similarity, the mind is induced to ascribe identity to the successive qualities of the wax. This is stage (1) of the four-stage reaction. I have said that the propensity to ascribe identity to related objects tricks the mind. Hume is explicit, in a related context one section earlier, that this propensity is based on a “deception” (202) or an “illusion” (200). This illusion leads to an outright mistake, taking for an identical object what are in fact diverse or distinct objects. There is a second way of considering the succession of sensible qualities. We can survey or compare the qualities before and after an interval during which they undergo considerable change, say, from a hard cube at one time to a molten mass at another. Since identity requires an uninterrupted and unchanging object, this way of viewing the qualities makes us recognize that the earlier and later qualities are not identical. This is stage (2). So we hold contradictory beliefs, ascribing both identity, at (1), and diversity, at (2), to the succession of qualities. As Hume says, there is “a kind of contrariety in our method of thinking” (T 220). We might expect that in these circumstances the mind corrects and thus gives up the belief that the earlier and later qualities are identical. After all, comparing the later sensible qualities with those that are more removed in time, we see that the belief in their identity is false. According to Hume, however, we do not simply relinquish the belief in identity. The psychological trick is fairly powerful. We remain in the grip of the propensity to ascribe identity to related objects even after we recognize that the objects are not identical. In the discussion of similar contradictions in the preceding section of the Treatise (T 205–6, 214–16), Hume tells us that such conflicts make us uneasy. Indeed, he portrays the contradictions as resulting in a psychological battle or war: the opposing beliefs are “enemies” which “attack” (215) each other in “combat” (205) and “struggle” in an effort to “destroy” each other (215; cf. 221). In other words, we suffer inner conflict; this causes “uneasiness” (205, 206; cf. 215), discomfort which we seek to relieve. This discomfort and the effort to relieve it constitute stage (3) of the reaction. Our recourse is to retreat from the belief in identity. We retreat minimally, only so far as is necessary to relieve the uneasiness. (One could imagine other principles of retreat, for example, trying to “split the difference” between opposing beliefs.) This retreat consists in supposing the existence of a wholly unobserved object which is unchanging as well as uninterrupted, and in which sensible qualities inhere. This is the final stage, stage (4), of the psychological reaction.

10. For an extended discussion of the operation of the propensity and its effects, see my 1991, esp. §§2–4.

hume’s explanations of meaningless beliefs

169

It is important that the belief in material substratum is not a genuine resolution of the conflict. Hume writes that the belief is simply “the means by which we endeavour to conceal [the contradictions]” (T 219, emphasis added). He writes of a similar supposition, that of a soul, that it functions “to disguise . . . variation” (254, emphasis added) in our perceptions. Postulating a substratum tends to obscure the contradiction to which it is addressed. Though Hume does not pause to explain why the supposition of a substratum fails to resolve the contradiction, it is not difficult to see what he has in mind. The propensity to ascribe identity to related objects is triggered by an observed succession of related sensible qualities, so that at stage (1) we are inclined to ascribe identity to the observed qualities. The supposition of a material substratum at (4) locates the identity in an “unknown and invisible” object (T 220), a wholly unobserved object. The mind retreats from the belief at (1) without entirely giving it up. The belief in the identity of the observed qualities survives in the form of belief in the identity of an unobserved substratum. Belief in a substratum goes some way toward satisfying the original inclination, by ascribing identity to something, but does not fully satisfy the inclination; belief in a substratum mislocates the identity that we are inclined to ascribe to observed objects, placing it instead in an unobserved object. For this reason, the supposition of a substratum is an unstable resolution of the contradiction. After his explanation of the belief in material substratum is complete, Hume still writes in a deprecating way about the concept of substratum, as an “unintelligible something” (T 220) and as “incomprehensible” (222; cf. 224). He is here calling attention to a defect in meaning or content.11 He does not show how the idea of material substratum is “derived,” in accordance with his Lockean theory of meaning, by approved psychological operations on ideas. The notion of material substratum is not, strictly, contentful or meaningful. Yet Hume has just purported to explain the belief in the existence of material substrata. This takes us back to the nasty puzzle: what is the content of the belief to be explained? I suggest that the mistake or illusion at stage (1) is transformed or transmuted, under the pressure of the conflict and uneasiness at (3), into a conceptual confusion. Once (3) is reached, the mind seeks relief. Unable to remove or resolve the conflict, we “endeavour to conceal” or “to disguise” the contradiction. The “concept” of material substratum, so far as we have it, emerges at the same stage, stage (4), as the belief in material substratum. We do not first form or acquire the concept and then the belief.12 Rather the concept and the belief are of a piece. What

11. Craig takes it that Hume started out “with a theory of the thinkable and then found himself happy to transform it into a theory of the knowable” (1987, 126–27), so that epistemic considerations supplant semantic ones. I do not think this does justice to the persistence of Hume’s expressions of misgivings about meaning. 12. In his treatment in I.iv.2 of another product of the propensity to ascribe identity to related objects, Hume provides an explanation of why our idea of the continued and independent

170

reflection and the stability of belief

I have called “quasi-content” arises together with the belief that obscures the contradiction at (4); the quasi-content and the belief result from the same set of psychological pressures. In thinking about the explanation of quasi-content which I am attributing to Hume, it is important to distinguish the mistake or illusion at stage (1) from the conceptual confusion at (4). The illusion at (1) consists in ascribing identity to related, but changing, observed sensible qualities. The conceptual confusion sets in at (4), when we ascribe identity to unobserved objects as best we can conceive them, thus postulating something “unintelligible” and “incomprehensible.” The confused conception of substratum at (4) has content-like features insofar as it is the by-product of a determinate illusion at (1). It might be objected that Hume is not entitled to the claim that stage (4) yields any form of content.13 As Hume views it, (4) goes some way to resolve the conflict to which it is addressed, requiring that it possesses a kind of content. How does he preclude the possibility that at (4) we enter into some psychological state that simply squelches the uneasiness, in the way a pill or a drug might squelch tension or discomfort? Such states would reduce uneasiness, though they lack any sort of content. I take Hume to be assuming that the outcome of the psychological reactions is belief, or something like belief; this is a datum to be explained. I suggest we have encountered an instance of a recurrent pattern of explanation in Hume. My general thesis is this: for Hume, quasi-content is the product of a retreat, under the pressure of a conflict and attendant uneasiness, from an illusion to a conceptual confusion required to possess a belief that obscures the conflict.14 Differences in quasi-content are a function of differences in the initial illusions which give rise to them. This thesis generates responses to our earlier puzzles. How can Hume consistently set out to explain the psychological causes of a belief which is strictly meaningless? The answer is that such a belief has a surrogate for meaning, quasi-content in the sense just explained. How can it be appropriate for Hume to give different explanations of a number of strictly meaningless beliefs? The answer is that the beliefs differ in quasi-content, owing to differences in the illusions at stage (1). Quasi-content is determined by the content of the initial illusion and associated contradiction, together with the principle that the mind retreats minimally from the initial belief.

existence of body also constitutes a belief (T 208–9). But his point is not that we first form the idea and later form the belief. 13. I owe this objection, and the response to it, to William Taschek. 14. David Pears finds “two separate thrusts in Hume’s strategy, one against the meaningfulness of his opponent’s views and the other against their credibility”; he takes Hume “to attribute equal importance to each of them” (1990, 10). When Pears is in this frame of mind, I think he is insufficiently sensitive to the tensions between the two thrusts. Also, though Pears allows that these thrusts are “closely interdependent” (46), I believe the interdependence tends to conform to the general schema outlined in my text.

hume’s explanations of meaningless beliefs

171

What are we to make of Hume’s perfunctory formulations and applications of his empiricism about meaning? I suggest the following picture. Hume’s destructive arguments to meaninglessness are merely intended to generate a presumption that a would-be concept is not strictly meaningful, not derived from experience in accordance with Lockean empiricism about meaning. Since Hume has in hand an explanation of the (quasi-) content of the concepts, these presumptive arguments can be brief. His overall argument is not complete until he supplements the presumptive argument with an explanation, with reference to a specific illusion, of how the relevant quasi-content arises. This places a double burden on those who think the target ideas more contentful than Hume allows: they need both to criticize the presumptive argument and to show that Hume’s explanation of the quasi-content is deficient.

3. Quasi-content and Belief My response to the various puzzles about meaning emphasizes that quasicontents are not derived from experience in accordance with Hume’s strictures on the origin of ideas that possess strict content. This leads to an important objection: if, as tradition in Hume interpretation has it, beliefs are lively ideas, and if quasi-contents are not ideas derived from simple impressions that copy experience, then we cannot have genuine belief in these cases. I have no objection to taking the psychological reactions to yield a state that falls short of genuine belief, quasi-belief. If one insists that, for Hume, beliefs are lively ideas, the best course is perhaps to identify quasi-contents with quasi-ideas, thus allowing for quasi-belief. It would be well, however, to locate an alternative to the admission of quasi-ideas. The pressure in the direction of saying that stage (4) of the psychological reaction yields quasi-belief arises from the interpretation that identifies beliefs with lively ideas. On this traditional interpretation, the propositional attitude of belief has a mental vehicle, lively ideas, whose constituents determine the content of the attitude. Lively ideas are thus “the language of thought” for belief. Applying this model to quasi-content, we have the problem of identifying a lively idea, or a constituent of a lively idea, which corresponds to a quasi-content. Hume’s Lockean empiricism about meaning, however, makes this problem intractable. My own view is that beliefs, for Hume, are not lively ideas, but rather (sufficiently) steady dispositions. A number of commentators have recognized dispositional strands in Hume’s account of belief.15 To quote just one

15. For example, Price credits Hume with “a hint or suggestion” of a dispositional analysis of belief (1969, 187; cf. 165, 188). Armstrong maintains that Hume “wavers” between dispositional and occurrent accounts of belief (1973, 71). Everson adopts a related “causal” and “functional” interpretation of Hume’s theory of belief (1988).

172

reflection and the stability of belief

passage from Hume, belief is “that act of the mind, which renders realities more present to us than fictions, causes them to weigh more in the thought, and gives them a superior influence on the passions and imagination” (T 629/1.3.7.7; cf. 624, 654, and EHU 49, 50–51, 56–57). Genuine beliefs are steady dispositions to display characteristic manifestations or typical effects on thought, the passions, and action—verbal and nonverbal actions, and conscious episodes, including lively ideas. Lively ideas are, for Hume, the occurrent thoughts in which (dispositional) beliefs are sometimes manifested. Lively ideas thus retain a role within the dispositional account of belief, giving the traditional interpretation its due. The dispositional account of belief makes room for quasi-beliefs associated with quasi-contents. Within the dispositional framework, lively ideas are the vehicle of occurrent thoughts which manifest (dispositional) belief, though not the vehicle of dispositional belief itself. Lively ideas are one, but only one, of the characteristic manifestations of genuine (dispositional) belief. Quasi-beliefs are steady dispositions to characteristic manifestations, apart from lively ideas. Quasi-beliefs have a variety of typical effects on thought, the passions, and action, but lively ideas are not among them. There is one element missing from my account to this point. Hume should allow that quasi-beliefs are sometimes manifested in occurrent thoughts, even though these thoughts cannot consist in lively ideas. It is here that we can press into service Hume’s remarks at Treatise 224, which I quoted (in part) in §1: For it being usual, after the frequent use of terms, which are really significant and intelligible, to omit the idea, which we wou’d express by them, and to preserve only the custom, by which we recal the idea at pleasure; so it naturally happens, that after the frequent use of terms, which are wholly insignificant and unintelligible, we fancy them . . . to have a secret meaning, which we might discover by reflection. Strictly meaningless terms serve as placeholders for ideas we do not possess. Hume can appeal to the subvocalization of terms or expressions (such as ‘material substratum’) that do not stand for ideas, together with other, meaningful, words. Such inner speech serves as an occurrent manifestation of quasi-belief. Furthermore, the expression ‘material substratum’, for example, is associated with the quasi-content at stage (4) in virtue of being employed in the context of the observations (of a succession of uninterrupted and gradually changing sensible qualities) that activate the propensity that gives rise to the illusion at (1). In this way, different expressions, ‘material substratum’, ‘immaterial substratum’, ‘external existence’, ‘necessary connection’, though not strictly meaningful, are associated, in both outer and inner speech, with different quasi-contents. Which quasi-content is in play depends on the context in which these expressions are placed or used. That explains how we can express different meaningless concepts and

hume’s explanations of meaningless beliefs

173

different meaningless beliefs. This completes a sketch of how quasi-content and quasi-belief fit into Hume’s theory of belief. I can only summarize the evidence for attributing a dispositional theory of belief to Hume. My position is that, quite apart from the issues discussed in this chapter, we must construe beliefs as steady dispositions.16 In the first place, this interpretation is required to explain how the results of poetical enthusiasm can exceed belief in vivacity (T 630–31/1.3.10.10), but amount only to “counterfeit belief” (123) or “the mere phantom of belief or persuasion” (630/1.3.10.10). In such cases the actions and internal episodes that are among the characteristic manifestations of genuine belief arise from sources other than steady dispositions, sources that mimic belief in some of their effects. These internal episodes include lively ideas. A dispositional interpretation is also required to accommodate Hume’s repeated statements that beliefs result from a process of infixing (86, 96, 225, 453, 624, 625, 629/1.3.7.7; cf. 99, 121) and that they are fast, firm, settled, solid, and steady (97, 105, 106, 108, 116, 121, 624, 625, 626, 627/1.3.5.4, 629/1.3.7.7, 631/1.3.10.10), rather than momentary, floating, and loose (97, 106, 110, 116, 123, 595, 624, 625). Such notions as “steadiness” apply more naturally to dispositions than to occurrent states.17

4. Additional Examples: Taste Conjoined in Place with Extension, and External Existence In the remainder of this chapter, I shall consider some additional examples of the psychological reaction that gives rise to quasi-content. Hume’s explanation of the belief in immaterial substrata or souls in Treatise I.iv.6 follows much the same pattern as that of material substrata. I turn, however, to consider two examples that might seem further afield. These will confirm my interpretation of quasi-content as the result of a four-stage psychological reaction; they will also play a role in a reconstruction of Hume’s treatment of the idea of necessary connection within my interpretative framework. I begin with Hume’s consideration of taste in I.iv.5. He claims that smells, sounds, and tastes do not have spatial location, but that they seem to be spatial (T 235–36). His example is the taste of a fig. He observes that various relations hold between the internal impression of taste and other qualities of the fig, such as its color and shape. The taste and the shape are regularly conjoined and contemporaneous. Hume maintains that “when objects are united by any relation, we have a strong propensity to add some new relation to them, in order to compleat the union” (237). In light of this 16. My interpretation of Hume’s account of belief, and how it differs from “counterfeit belief,” has been influenced by that of MacNabb (1951/1966, esp. 71–81). [For a textual note on the citation from Treatise 123 in this paragraph, see “References to Hume,” this volume.] 17. I develop these arguments for taking belief in Hume to consist in a steady disposition in my 1995a, esp. 122–25 [this volume, ch. 7, §4], and my 2001b [this volume, ch. 5].

174

reflection and the stability of belief

propensity, we “endeavour to give [the taste and shape] a new relation, viz. that of a conjunction in place.” This yields a stage (1), the supposition that the taste is conjoined in place to the fig. This is a mistake or an “illusion,” due to the propensity to add a new relation to related objects. Indeed, reflection rushes in to uncover a dilemma in (1). Does the taste exist only in some part of the fig, or in every part? The taste cannot exist only in some part of the fig, for every part has the same taste; and the taste cannot exist in every part of the fig, for in that case, Hume says, the taste would be extended (cf. T 238). The belief at (1) is thus meaningful but internally inconsistent. These reflections give rise to a stage (2), the belief that the taste is not conjoined in place with the fig’s extension. The contradiction at (1) and (2) leads to stage (3): “Being divided betwixt these opposite principles, we renounce neither one nor the other” (T 238). Here we are genuinely “divided” or torn. We cannot simply say that the taste is conjoined with the extension, for then the contradiction will be obvious. The propensity to add new relations to related objects nevertheless has a strong pull on us. So we retreat from (1), just far enough to obscure the contradiction. This effort to relieve the uneasiness results in a stage (4): we suppose that the taste exists in every part of the fig and that it does so in its entirety. Every part of the fig has the same taste, but since the taste exists in its entirety in every part of the fig, the taste is not extended after all. Hume observes that the belief at (4) is conceptually confused: it amounts to saying “that a thing is in a certain place, and yet is not there” (238). The illusion at stage (1), that the taste is conjoined in place with extension, is distinct from the claim at (4) that the taste in its entirety is conjoined with every part of the fig. The belief at (1) has sufficient meaning for us to be able to expose its incoherence. Under the pressure of the conflict and uneasiness, the illusion at (1) is transformed into a conceptual confusion. The effect of (4) is to conceal the conflict, Hume says, precisely because we “involve the subject in such confusion and obscurity, that we no longer perceive the opposition” (T 238) between stages (1) and (2). The example of taste follows the same pattern as that of material substratum. The difference is that in the case of material substratum, it is the propensity to ascribe identity to related objects that gives rise to the initial illusion; in the case of taste, it is the propensity to add new relations to objects united by another relation. This is the feature of the example that will be important later in my reconstruction of Hume’s treatment of the idea of necessary connection. My discussion of necessary connection will also rely on one further example of the four-stage reaction, Hume’s explanation of the idea of external existence. He writes in I.i.6: “since nothing is ever present to the mind but perceptions, and since all ideas are deriv’d from something antecedently present to the mind; it follows, that ’tis impossible for us so much as to conceive or form an idea of any thing specifically different from ideas and impressions” (T 67; cf. 188, 216, 218, 241). Yet Hume also sets out to

hume’s explanations of meaningless beliefs

175

explain the origin of the belief in external existence. Here again we confront the nasty problem: if the belief has no content, what is there to explain?18 Hume’s explanation of the belief in external existence begins with the common person’s belief that the objects that are immediately perceived have a continued existence when not present to the senses and an existence independent of the mind and perception (T 188). Philosophers distinguish external objects from internal perceptions. The common person, according to Hume, does not make this distinction, and believes that the objects that are immediately perceived, objects that are in fact internal perceptions, have a continued and independent existence (202). The common person’s belief in the continued and independent existence of what are in fact internal perceptions arises as follows. When Hume perceives a mountain, or furniture in his chamber (T 194–95), the perceptions before and after he shuts his eyes or turns his head are related in that they are unchanging, though interrupted (204). This relation is similar to that of being unchanging and uninterrupted. In light of this similarity, the propensity to ascribe identity to related objects (the operative propensity in the case of substratum) tricks the mind into ascribing identity to the successive perceptions. Identity, however, requires unchanging and uninterrupted existence. The mind therefore supposes that the perceptions of the mountain or furniture are not interrupted. Here is a stage (1), the supposition that our perceptions have a continued existence during the interruptions when they are not perceived. This is an “illusion” (200), produced by the propensity to ascribe identity to related objects (cf. 199, 205–10). The belief at (1) is about the continued existence of perceptions, objects internal to the mind. Our interest is in the idea of external existence. Hume claims that “a very little reflection and philosophy” (T 210; cf. 214, 215) leads us to notice the mistake at (1). For example, if we press one eyeball, “we immediately perceive all the objects to become double, and one half of them to be remov’d from their common and natural position” (210). Such experiences show, Hume thinks, that perceptions are dependent on the mind (cf. 210–11, 214, 215). This yields a stage (2), the belief that perceptions do not have a continued existence when not perceived. Hume is explicit that the contradiction results in conflict and uneasiness (cf. T 214–16). This is a stage (3). But the philosopher does not simply sacrifice the belief at (1). The philosopher, as well as the common person, is in the psychological grip of the propensity that leads to this belief (cf. 215). Philosophers accordingly retreat to a stage (4). In order to relieve the

18. Recent interpretations that attribute to Hume the position that it is meaningful to suppose that external objects and/or necessary connections exist, though we do not know whether they exist, or that Hume believes in the existence of external objects and/or necessary connection, must show that he makes room for at least the meaningfulness of these beliefs. Thus, for example, G. Strawson calls attention to an apparent “meaning tension” in Hume. See his 1992, 120–22, and other index entries at 290. For an excellent critical review of this literature, see Winkler 1992, esp. §2.

176

reflection and the stability of belief

discomfort, they suppose “the double existence of perceptions and objects” (211, 215). Philosophers distinguish between internal perceptions and external objects, and suppose that although perceptions have an interrupted existence, external objects have an uninterrupted existence (211). They also suppose that external objects cause (217) and resemble (213, 216–17) internal perceptions: “The relation of cause and effect determines us to join the other of resemblance” (217). (There is a lacuna here in the sense that Hume never provides an explicit psychological account of the specifically causal aspect of the hypothesis of double existence. It is an interesting question why he does not do so.) Philosophers thus come to embrace the “philosophical system” (212, 213), indirect or representative realism. This supposition of the existence of external objects is “only a palliative remedy” (T 211; cf. 215), a “pretext” (216) by which we “elude” (215) the conflict. By ascribing continued existence to some objects, albeit ones that have not been observed, the supposition of double existence goes some way to satisfy the inclination to ascribe identity to the unchanging though interrupted perceptions. On the other hand, belief in external objects mislocates the identity that we are initially inclined to ascribe to observed objects, placing it instead in objects that are unobserved. Hume goes too far when he writes that “each [of nature and reason] may find something, that has all the conditions it desires” (215, emphasis added). The supposition is a symptom, rather than a resolution, of a conflict that persists even after stage (4) has been reached. My position should give no comfort to commentators who maintain that Hume thinks the idea of external existence tolerably meaningful. These commentators contend that Hume admits “external existence” as a “relative” idea. We have meaningful concepts of internal perceptions and we have a meaningful concept of causation; external existence is whatever it is that causes our internal perceptions.19 I do not think this reading consistent with the tenor of a number of passages where Hume writes that external existence, as specifically different from perceptions, is inconceivable or incomprehensible (cf. T 67–68, 188, 241), the conception of “a relation without a relative” (241).20 More fundamentally, Hume’s treatment of external existence parallels his treatments of substratum and of the local conjunction of taste with matter. In each case, the relevant “concept” arises in the service of concealing or obscuring an underlying conflict rooted in an initial illusion or deception. In each case, conflict and uneasiness transmute an illusion into a confused quasi-content inherent in a supposition that provides a confused resolution of the contradiction. This is not a sympathetic account of the origin of the 19. See, e.g., Livingston 1984, 80–81, 155–57; Flage 1990, 37–38, 39, 42–51; and G. Strawson 1992, 49–58, and the other index entries for “‘relative’ ideas” at 291. For critical discussion of this sort of view, see Blackburn 1990; Pears 1990, 194–96; Winkler 1992, esp. §2; and Broackes 1993, esp. 104–7, 110. 20. For a more extended discussion of these passages, see Blackburn 1990, 239–41.

hume’s explanations of meaningless beliefs

177

idea of external existence. This is the feature of Hume’s explanation of this idea which will be important later in the chapter.

5. Necessary Connection I turn to Hume’s treatment of the idea of necessary connection. I take the Treatise and the first Enquiry to represent substantially the same view, though the version in the Enquiry is diluted in some respects. I shall state, without defense, some of my interpretative assumptions. First, Hume holds that causation consists in regularity or constant conjunction. Of course, he famously observes: “There is a necessary connexion to be taken into consideration” (T 77). If the small pot of water has been on the high flame for fifteen minutes, it must be boiling. Second, Hume has no objection to identifying necessary connection in the objects with the constant conjunction of instances of the relevant types of objects. His first definition of ‘necessity’ does precisely that: necessity “consists . . . in the constant conjunction of like objects” (EHU 97; cf. T 409). But there is no necessary connection in the objects, in the causes and effects, over and above their constant conjunction. Third, though we have a conception of necessary connection insofar as it is identified with constant conjunction, we have no legitimate conception of necessary connection in objects, as distinct from constant conjunction. Hume begins both the Treatise and first Enquiry sections on necessary connection with statements of his Lockean theory of meaningfulness (T 155; EHU 61–62).21 He subsequently writes in the Treatise discussion: “when we speak of a necessary connexion betwixt objects, and suppose, that this connexion depends on an efficacy or energy . . . we have really no distinct meaning, and make use only of common words, without any clear and determinate ideas” (T 162; cf. 161). Similarly, in the first Enquiry, he writes: “we cannot . . . point out that circumstance in the cause, which gives it a connexion with its effect. We have no idea of this connexion, nor even any distinct notion what it is we desire to know, when we endeavour at a conception of it” (EHU 77; cf. 74, 82). These passages strongly suggest that the conception of necessary connection (in objects, over and above constant conjunction) is defective with respect to meaning. My final assumption is that Hume nevertheless wants to explain our inclination to believe that there is a “mustness” or “necessity,” a necessary connection (in objects, over and above constant conjunction). At the same time, that we have no idea of such a connection and that the relevant expressions are meaningless are important strands in Hume’s thinking. Here we have a new instance of what I have called “the nastier problem” for Hume’s deployment of his Lockean theory of meaning. As Stroud writes, “If we can

21. Winkler 1992, 555–56.

178

reflection and the stability of belief

have no idea of necessity as something residing in objects, and our only idea of it is as something that occurs or exists in the mind, then we cannot even have the false belief that necessity is something that is objectively true of the connections between objects or events in our experience. To have that false belief we need at least an idea of necessity as something true of the connections between events.”22 To show how this problem plays out in Hume’s texts, I shall consider some basic ingredients of his explanation of the belief in necessary connection. Suppose we have repeatedly observed a constant conjunction, say, fire followed by smoke. When we next observe fire, we feel a mental determination to expect smoke to follow. This feeling of the determination to expect smoke is the result of the repeated observation of the conjunction, of conditioning—Hume’s “habit” or “custom.” Hume writes that “we transfer the determination of the thought to external objects, and suppose [a] real intelligible connection betwixt them” (T 168, emphasis added; cf. EHU 77–78n.). The mind cannot literally move the felt determination to objects. Hume means that the mind is induced, at least, to believe that the internal impression resides in external objects. He holds that this belief is a mistake: the felt determination is “a quality, which can only belong to the mind,” so that when we transfer the felt determination to the causes and effects “we make the terms of power and efficacy signify something . . . which is incompatible with those objects” (T 168). As it stands, this is an explanation of why we have the mistaken, but meaningful, belief in necessary connection between causes and effects. We have an idea of necessary connection, namely, the felt determination of the mind to expect one object to follow another. We transfer this idea to objects, supposing a felt determination between them. In this we are simply mistaken. If this is Hume’s explanation, what has happened to his theme that the very conception of necessary connection (in objects, over and above constant conjunction) is defective in meaning? On the explanation Hume has offered, the belief in necessary connection is false, but not meaningless. So what is all the fuss about defects in meaning? On the other hand, how could Hume explain why we believe in necessary connection and also claim that we have no legitimate conception of it? This problem has a now familiar structure. Evidently the conception of necessary connection must be defective with respect to meaning, but also must have sufficient content to permit an explanation of the origin of the belief in its existence. This explains some of Hume’s qualifications in summary statements about the idea of necessary connection. For example, “we deceive ourselves, when we imagine we are possest of any idea of this kind, after the manner we commonly understand it” (T 161, emphasis added); “the necessary conclusion seems to be that we have no idea of connexion or power at all, and that these words are absolutely without any meaning, when employed either in philosophical reasoning

22. Stroud 1977, 83.

hume’s explanations of meaningless beliefs

179

or common life” (EHU 74). His point is not that there is some alternative way to understand the expression which affords it strict meaning, but rather that it has quasi-content. If this is right, we need to identify, in the context of necessary connection, a relevant instance of Hume’s pattern of explanation for quasi-content. In order to secure this result, I offer an extension of his explanation of the belief in necessary connection. He writes: [T]he mind has a great propensity to spread itself on external objects, and to conjoin with them any internal impressions, which they occasion, and which always make their appearance at the same time that these objects discover themselves to the senses. Thus as certain sounds and smells are always found to attend certain visible objects, we naturally imagine a conjunction, even in place, betwixt the objects and qualities, tho’ the qualities be of such a nature as to admit of no such conjunction, and really exist no where. But of this more fully hereafter. (T 167) The mind’s transferring the felt determination to external objects is an instance of the mind’s propensity to spread itself on external objects, that is, to conjoin with external objects internal impressions which they occasion. Hume cites another instance: conjoining with external objects sounds and smells which they occasion. He writes that he will discuss this more fully later and provides a footnote reference to the section that contains the discussion of taste, material that also applies to sounds and smells (235–36). Even in the watered-down treatment in the first Enquiry he has this analogy in view, observing that “nothing is more usual than to apply to external bodies every internal sensation, which they occasion” (EHU 78n.). The idea of necessary connection, so far as we have it, results from the same propensity as the idea of taste conjoined in place with extension. In the discussion of taste, the mind’s propensity to spread itself on external objects is itself explained with reference to the propensity to add a new relation to objects that bear another relation to each other. The mind’s propensity to transfer felt determination to external objects thus results from the more general propensity to add a new relation to related objects. I shall follow Hume’s lead and work through the analogy he asserts between taste and necessary connection. In the case of taste, the internal impression is conjoined and contemporaneous with a single object, the fig. We add a new relation, conjunction in place with that object. In the case of the felt determination of the mind to expect smoke, the internal impression occurs just after the observation of the fire and just before the observation of smoke. (I have in view cases where the felt determination occurs before the observation of an observed object’s effect. Even if this condition is not satisfied in all cases, the existence of sufficient cases of this sort explains why we locate the felt determination between the objects in time, thus construing it as a connection, tie, or bond.) Since the felt determination is conjoined with both the fire and the smoke, the propensity to add a new relation to

180

reflection and the stability of belief

related objects will induce us to transfer the internal impression to the objects outside us and, so far as possible, to place it temporally between them. This account fits neatly with Hume’s emphasis on a search for a connection, tie, or bond “between” (EHU 70, 74) or “betwixt” (T 162, 163, 166, 168) the causes and effects. This gives us a stage (1): we believe that the felt determination exists between the objects. This belief is obviously mistaken. Hume writes that it is an “illusion” (T 237) to suppose that an internal impression exists outside the mind. He also writes of “illusions” (267) of the imagination, in connection with remarks about necessary connection in I.iv.7. Hence we arrive at stage (2): we believe that there is no felt determination between the objects. The resources for this stage are explicit in Hume’s remark (cited previously) that necessary connection is “a quality, which can only belong to the mind,” a quality “incompatible” with the objects to which it is applied. There is a contradiction between (1) and (2), and hence we can expect a stage (3), conflict and uneasiness, which the mind will endeavor to resolve. Though Hume does not write of conflict and uneasiness in his sections on necessary connection, he does say that when we transfer the internal impression to external objects, “obscurity and error begin then to take place” (T 168, emphasis added). We have seen that obscurity is the hallmark of a retreat from an illusion and attendant contradiction. What is the vehicle of retreat from stage (1)? One version consists in a stage (4a): we believe that there exists something between the objects that resembles the internal impression of felt determination. The conception of something in or between the objects that resembles the internal impression is confused. We have no idea what that could be like. It is difficult to see what could resemble the felt determination, except the felt determination itself. Perhaps the idea of the felt determination copies and resembles the impression, but an idea is not a candidate for a necessary connection between objects. As in other cases, (4a) functions to conceal a conflict. We are thinking of something, we know not what, related (by resemblance) to something we experience. The illusion at (1) is thus transmuted, under pressure of the conflict and uneasiness, into a confused conception, in the service of obscuring the contradiction. The idea of necessary connection turns out to be a confused quasi-concept, on a par with substratum and the local conjunction of taste with extension. Might there be a second avenue of retreat, one that puts the idea of necessary connection in a more favorable light? The most promising candidate is (4b): we believe that there exists something between the objects that causes or produces the internal impression of felt determination. Of course on Hume’s account of causation there is no such thing. The felt determination is caused by repeated observations of conjoined objects; there is nothing between the objects, between the fire and smoke on a given occasion, that causes the internal impression. The issue, however, is whether we have reached a legitimate conception of necessary connection, rather than a confused quasi-concept.

hume’s explanations of meaningless beliefs

181

I believe that Hume would reject the thought that (4b) gives rise to a concept that is meaningful, strictly speaking. In the first place, there is the question of how we are to understand the concept of “causation” in (4b). If (4b) figures in the explanation of the origin of a concept of necessary connection (over and above constant conjunction), it cannot rely on that concept, on pain of circularity. There is the austere concept of constant conjunction, though it seems odd for the mind prereflectively to help itself to an understanding of causation that emerges from philosophical and psychological analysis.23 In the second place, Hume would regard (4b) as unstable. In discussing the “philosophical system,” or indirect realism, he tells us that our adding resemblance to objects related by causation (T 213, 216–17) results from the “strong propensity to compleat every union by joining new relations to those which we have before observ’d betwixt any ideas” (217). This is the propensity that leads to the belief at (1) that the felt determination exists between the objects in the first place. Even if we somehow form the supposition (4b), the propensity to add new relations to related objects will lead us to add resemblance to causation, yielding (4c): the belief that there exists something between the objects that causes and resembles the internal impression of felt determination. This version of (4) inherits the confusion inherent in (4a). Even putting these worries aside, there is another objection to construing (4b) as providing a legitimate conception of necessary connection. It is here that my earlier discussion of the idea of external existence comes into play. The idea of external existence, insofar as we have it, is the idea of whatever it is that has an uninterrupted existence and causes our internal perceptions. This is similar in form to the idea of whatever it is between external objects that causes the internal impression of felt determination. Both are “relative” ideas, with a would-be object characterized in terms of a relation, causation, to something we experience. Indeed, it is this feature of the ideas that invites the thought that they are legitimate. I have argued, however, that Hume assigns the relative idea of external existence the same status with respect to meaning as he does the concepts of substratum or local conjunction of taste (in its entirety) with extension. All these concepts have the same genesis: they arise in order to obscure conflict and uneasiness due to an initial illusion and the associated contradiction. If I am right that Hume regards the relative idea of external existence as a confused quasi-concept, the relative idea of necessary connection along the lines of (4b) should suffer the same fate. A treatment of the idea of necessary connection as arising by the route from (1) to (4b) is every bit as unkind as the parallel treatment of the idea of external existence. Whether our minds rely on (4a) or (4b), we land in a quasi-concept.

23. The point here is similar to the criticism of G. Strawson advanced by Winkler 1992, 558–59, and Broackes 1993, 107.

182

reflection and the stability of belief

I conclude that we can attribute to Hume views about quasi-content that apply to the idea of necessary connection, as well as substratum, the local conjunction of taste in its entirety with extension, and external existence. In these cases, the target ideas can be viewed as confused quasi-concepts, arising from an initial illusion under the pressure of conflict and uneasiness. This psychological reaction explains both the quasi-content and the quasibelief. Hume’s meaning-empiricism is thus two-pronged. The first prong is a presumptive argument against strict content, ideas derived from perceptions in accordance with Hume’s Lockean theory of meaning. This is combined with an explanation to account for surrogate meaning or quasi-content. This interpretation explains the perfunctory character of Hume’s Lockean arguments to meaninglessness. It also explains how Hume can consistently provide psychological explanations of a number of different beliefs which do not meet Lockean standards for strict meaning.24

24. I have benefited greatly from reading versions of this chapter to audiences at Union College, Wellesley College, and the Ohio State University Conference on the History of Modern Philosophy and from discussion at a meeting of the Propositional Attitudes Task Force at Smith College. I am indebted to Michael Jacovides and Kenneth Winkler for detailed written comments; to Diana Raffman and William Taschek for pressing difficulties in my account of quasicontent; to David Velleman for helping me to clarify and develop my views about quasi-belief; and to an anonymous referee for helpful suggestions. I wrote this chapter during a year as Fellow, Center for the Study of Modern Philosophy, University of Massachusetts, Amherst. I am grateful to the Department of Philosophy, University of Massachusetts, for support, and to the University of Michigan for sabbatical leave.

hume’s explanations of meaningless beliefs

183

7 Hume on Stability, Justification, and Unphilosophical Probability

1. Introduction a number of passages in the Treatise suggest that Hume thought in terms of a particular picture of the conditions under which beliefs are justified, or reasonable.1 Hume did not develop the details of this picture, so that it constitutes more of an epistemological orientation than a systematic theory. For ease of exposition, however, I often write of “Hume’s theory of justification” in referring to the way of thinking about justification that emerges on my reconstruction of Hume’s position. In this chapter, I argue for two general theses about Hume’s theory of justification. One thesis is that Hume develops the theory in two stages, with opposite results: first, in order to ratify the theory as explaining or systematizing pretheoretical distinctions between justified and unjustified belief and, second, to show that these distinctions ultimately cannot be sustained with reference to the theory. The second thesis is that Hume tends to explain justification in terms of psychological stability. Although these theses are independent, the evidence for them is interconnected, so that I discuss them in a single chapter. Hume’s leading idea is that justification is to be explicated with reference to the psychological equilibrium of sets of doxastic states—sets of beliefs, inclinations to hold beliefs, and “quasi-beliefs” (§4). Hume, I believe, prizes stability in doxastic states. Stability is a dispositional property, the tendency of a doxastic state to remain in place, not to change. Two broad types of instability are of interest to Hume. First, there is the instability of a doxastic 1. [In this chapter, quotations of Hume are based on an electronic edition prepared by Thomas Beauchamp, David Norton, and Alexander Stewart (BNS).]

184

state owing exclusively to characteristics of the mechanism that produces or sustains the state. For example, some doxastic states have a tendency to change abruptly; in Hume’s terminology, they lack fixity. (Fixity, according to Hume, results from perception, memory, or a mechanism involving repetition.) Other doxastic states are infixed, but have a tendency to change gradually because of associationist degradation or decay. In both cases, the tendency to change is entirely a result of the mechanism that produces or sustains the doxastic state. Second, there is the instability of a doxastic state owing to disequilibrium in the set of doxastic states of which it is a member. A set of doxastic states is in disequilibrium insofar as one is inclined to revise the set solely as the result of consideration of the content of the doxastic states themselves. Even a doxastic state that is infixed can suffer from disequilibrium. Similarly, a doxastic state that is unstable because it is not infixed, or because it degrades, can have a further instability if it is a member of a set of doxastic states in disequilibrium. Hume thinks that a doxastic state that suffers from lack of fixity does not deserve to be called a belief (§4). And he thinks, roughly, that a doxastic state (whether or not it qualifies as a belief) that suffers from disequilibrium does not deserve to be called justified. I say “roughly,” because Hume prefers to evaluate the justificatory status of a belief with reference to generic properties of the belief-forming mechanisms that produce the belief, so that a doxastic state is justified just in case it results from mechanisms that tend to produce sets of doxastic states that are in equilibrium.2 It is a picture of justification along these lines that is developed in two stages. The first stage is constructive. Hume draws some pretheoretical distinctions between justified and unjustified beliefs. He invokes his theory of justification, together with some claims about the properties of relevant belief-forming mechanisms, in order to explain the pretheoretical distinctions. He sees himself as having some success in doing so, so that the theory of justification is ratified as capturing the pretheoretical distinctions. The second stage is destructive. Hume alleges that the theory of justification, combined with some additional properties of our beliefforming mechanisms, has the consequence that no belief is justified, and hence that the pretheoretical distinctions cannot be sustained with reference to the theory after all. In the face of this conclusion, Hume opts to maintain the theory of justification, and to abandon the pretheoretical distinctions. As we shall see, he offers ambivalent and even contradictory assessments of the tenability of the pretheoretical normative distinctions that he draws. The idea that Hume develops his theory of justification in two stages provides an explanation of the tensions in his attitude toward these distinctions. 2. This is a modification of my earlier view that, for Hume, cognitive faculties are evaluated with reference to their contribution to doxastic permanence. Impermanence is not objectionable insofar as it results from the acquisition of new evidence, rather than from disequilibrium. See my 1990 [this volume, ch. 2], §7.

hume on stability, justification, and unphilosophical probability

185

My project in this chapter is to construct a cumulative case for attributing to Hume a stability-based picture of justification, not to develop the picture into a fully articulated theory. In §2, I provide some evidence that Hume seeks to explicate justification in terms of equilibrium. This evidence is based, in part, on texts that I take to constitute the destructive stage of Hume’s discussion of the theory. In these passages, culminating at Treatise I.iv.7, Hume brings himself to conclude that no belief is justified. I argue that the attribution of a stability-based theory of justification to Hume provides a promising explanation of this development. Hume comes to the conclusion that no belief is justified on the ground that our belief-forming mechanisms inevitably tend to produce sets of beliefs that are in disequilibrium. In §§3–4, I consider the bearing of Hume’s theory of justification on his discussion at I.iii.13 of four species of unphilosophical probability. This provides a case study—or a set of interrelated case studies—of the fruitfulness of my interpretive framework. Hume’s treatment of unphilosophical probability belongs, in large measure, to the constructive stage of the development of his theory of justification. On my interpretation, unphilosophical probabilities derive from disequilibria that arise within memory and causal inference, and that are eliminable. (One upshot of the interpretation is a thesis about Hume’s appeals to “general rules”; general rules are often second-order beliefs that serve either to destabilize, or to stabilize, systems of first-order beliefs.) Unfortunately, the inescapable instabilities introduced at I.iv.7 are waiting in the wings, and in the end Hume claims that their existence ruins his attempt to sustain the pretheoretical distinctions about justification.

2. Stability and Hume’s Commitment to Epistemological Distinctions I begin with an inventory of epistemological distinctions that Hume endorses. Hume draws these distinctions by enumeration, with reference to specified kinds of belief-forming mechanisms. Reason, or the understanding, comprises demonstrative and probable reasoning. Probable reasoning admits of degrees of evidence, and comprises proofs and probabilities. Probability in turn divides into the probability of chances, of causes, and inferences from analogy. Reason is contrasted with the imagination, in a narrow sense of the term. Taken in this sense, the imagination comprises ideas fainter than memory, and generated by an associative process other than those constituting demonstrative and probable reasoning. Hume’s enumeration of these distinctions is “pretheoretical” in that it does not provide a theoretical account of what the specified kinds of inferences have in common that demarcates them from those that are not justified. Let us see in somewhat more detail how the epistemological distinctions in our inventory unfold in the Treatise. Hume claims that all assurance or belief in matters of fact, that is not based solely on present perception or on

186

reflection and the stability of belief

memory of what has been perceived, is based on causal inference, inference about objects causally related to objects that have been perceived (T 73–74, 82–83, 84, 87, 89, 107–8, 198, 212). (In discussing Hume’s causal theory of assurance, I use ‘inference’ to refer to any associative transition from one idea to another.) According to Hume, causal inference is “just” or justified: “The only connexion or relation of objects, which can lead us beyond the immediate impressions of our memory and senses, is that of cause and effect; and that because ’tis the only one, on which we can found a just inference from one object to another” (89). This passage is contained in the paragraph bridging Selby-Bigge pages 89 and 90, the paragraph that completes Hume’s initial formulation of the problem of induction. At page 88, Hume begins his inquiry into the nature of the transition from belief in matters of fact based on present perception or memory, to the belief in an unobserved cause or effect. Such an inference might seem to presuppose that nature is uniform, but belief in the uniformity of nature cannot be founded on any argument (88–90). We think of Hume’s problem of induction as intended to show that there can be no justification for believing the conclusion of any nondemonstrative argument. It is therefore surprising to find, in the paragraph bridging pages 89 and 90, the claim that causal inference is just. This claim is no lapse. At pages 107–8 of I.iii.9 there is an extended restatement of the theory that all belief in matters of fact is based on perception, memory, and causal inference. Hume describes two systems of beliefs or “realities”: the first is based on the senses and memory alone; the second system supplements the first, and is based on custom, or the relation of cause and effect. Hume attributes the second system to “the judgment.” He claims that beliefs not founded on perception, memory, and the relation of cause and effect—beliefs not included in either system of realities—“are merely the offspring of the imagination.”3 In a subsequent footnote, Hume distinguishes two senses of the term ‘imagination’: [A]s our assent to all probable reasoning is founded on the vivacity of ideas, it resembles many of those whimsies and prejudices which are rejected under the opprobrious character of being the offspring of the imagination. By this expression it appears that the word, imagination, is commonly us’d in two different senses; and tho’ nothing be more contrary to true philosophy, than this inaccuracy, yet in the following reasonings I have been oblig’d to fall into it. When I oppose the imagination to the memory, I mean the faculty, by which we form our fainter ideas. When I oppose it to reason, I mean the same faculty, excluding only our demonstrative and probable reasonings. (T 117–18n.)

3. Surprisingly little attention is given to Hume’s seeking to assign causal inference normative pride of place. For discussions that are clear on this point, see Immerwahr 1977, esp. 58–63; Passmore 1952/1968, 60–63; and Price 1969, Lecture 7, esp. 173–75.

hume on stability, justification, and unphilosophical probability

187

I use subscripts to flag the two senses. Hume identifies the imagination in a wide sense, the imaginationw, with the faculty by which we form ideas fainter than memory. He identifies “demonstrative and probable reasonings” with “reason,” or “the understanding” (371n.).4 He identifies the imagination in the narrow sense, the imaginationn, with the imaginationw, exclusive of reason. The discussions of the two systems of realities and the two senses of “imagination” are located ten pages apart within a single section of the Treatise. When Hume writes at page 108 that beliefs not based on perception, memory, or the relation of cause and effect “are merely the offspring of the imagination,” he has in mind the imaginationn—otherwise there would be no room for his contrast with “the judgment.” Since beliefs based on the relation of cause and effect are not due to the imaginationn, they must be due to reason. The distinction between reason and the imaginationn, drawn against the background of the discussion of the two systems of belief at pages 107–8, captures epistemological distinctions Hume wishes to draw.5 The normative distinction between reason and the imaginationn generalizes Hume’s commitment to the justificatory status of causal inference; it is “reason,” demonstrative and probable reasoning, that is justified. Hume draws a variety of epistemic distinctions within “probable reasoning.” At I.iii.11, he divides nondemonstrative or probable reasoning into “proofs” and “probabilities.” Proofs are causal inferences conditioned by frequently observed constant conjunctions, where the present perception exactly resembles the observed instances (T 124, 130–31, 153). These “arguments from causation exceed probability, and may be receiv’d as a superior kind of evidence”; they “are entirely free from doubt and uncertainty” (124). (Such arguments are “just” at page 144.) Hume states that the distinction between proofs and probabilities serves not only “to preserve the common signification of words” but also to “mark the several degrees of evidence.” Hume shares the “common” (124) view that some arguments from causation— the ones he designates “proofs”—constitute a superior degree of evidence. Hume proceeds to examine “probability” at I.iii.11–12.6 At I.iii.13, he contrasts “proofs” and “probable reasoning” with “unphilosophical probability.” The claim that there are degrees of evidence is reiterated in the penultimate paragraph of I.iii.13 (153–54). It is standardly thought that the force of the problem of induction is to show that all nondemonstrative arguments have the same evidential force—none at all. Subsequent to his formulation 4. [For a textual note on 371n., see “References to Hume,” this volume.] 5. In the note at 117–18, Hume takes issue with using a word in two different senses as “contrary to true philosophy,” but he does not take issue with the negative assessment of beliefs based on whimsy and prejudice. 6. In I.iii.9, beliefs based on the relation of cause and effect are due to reason, the faculty of demonstrative and probable reasoning. Causal inference is not demonstrative, so that it must be a kind of probable reasoning. At I.iii.11, some causal inferences—the proofs—exceed probability. This discrepancy is purely terminological. At I.iii.9, Hume writes of “probable reasonings” in a wide sense in contrast to demonstrative reasoning. In I.iii.11, Hume’s distinction between proofs and probability is drawn within probable reasoning in the wide sense.

188

reflection and the stability of belief

of the problem of induction, however, Hume is clear that there are degrees of evidence. A complete reversal in Hume’s expressed attitude toward the epistemological distinctions he has drawn occurs between I.iv.4 and I.iv.7, the “Conclusion of this book.” At pages 225–26 of I.iv.4, Hume again asserts that causal inference is just: “One who concludes somebody to be near him, when he hears an articulate voice in the dark, reasons justly and naturally; tho’ that conclusion be deriv’d from nothing but custom.” By contrast, beliefs based on superstition are natural, but not just. Yet Hume writes at I.iv.7: “I am ready to reject all belief and reasoning, and can look upon no opinion even as more probable or likely than another” (T 268–69). This does not mean that Hume is ready to suspend all belief; some beliefs are irresistible (31, 225; cf. 128, 147). It means that Hume is ready to reject all belief as unjustified, even though some beliefs are irresistible. He is ready to do so because he can no longer accept judgments that draw distinctions with respect to the justificatory status of different beliefs. We have seen that Hume has committed himself to a variety of epistemic distinctions in previous sections of the Treatise. Prior to I.iv.7, he has held that causal inference is justified, that there is an epistemic distinction between reason and the imaginationn, and that there are degrees of evidence within nondemonstrative knowledge. At pages 268–69, Hume “can look upon no opinion even as more probable or likely than another”—Hume is ready to renounce all these distinctions. How are we to account for Hume’s contradictory assessments of the tenability of the epistemological distinctions he draws? I propose that the normative distinctions that Hume initially endorses, distinctions drawn by enumeration, reflect Hume’s pretheoretical epistemological commitments. Hume attempts to show that his theory of justification explains these commitments and sees himself as meeting with considerable success. The constructive stage is prominent as late as Treatise 225–26. The results of this stage, however, prove provisional; Hume comes to conclude that the normative distinctions to which he is pretheoretically committed ultimately cannot be sustained with reference to his theory of justification. This destructive stage culminates in I.iv.7. This two-stage model of Hume’s development of his theory provides a framework for thinking about the reversal in Hume’s expressed attitude toward epistemological distinctions. In order to flesh out these suggestions, we need to extract from the Treatise a theory of justification that Hume, on the one hand, intends to sustain his pretheoretical epistemological commitments and, on the other hand, views as failing to do so. If my interpretation is correct, we should expect that Hume is ready to reject all belief as unjustified because he comes to the view that equilibrium cannot be achieved. The considerations advanced at I.iv.7 that lead to Hume’s readiness to abandon all epistemic distinctions substantially fulfill this expectation. An initial consideration draws on a conclusion at I.iv.4, “Of the modern philosophy.” Hume writes in the final paragraph of that section: “there is a

hume on stability, justification, and unphilosophical probability

189

direct and total opposition betwixt our reason and our senses; or more properly speaking, betwixt those conclusions we form from cause and effect, and those that persuade us of the continu’d and independent existence of body” (T 231). At page 193, Hume attributes the belief in the continued and independent existence of body to the imagination rather than to the understanding, that is, to the imaginationn. At page 231, therefore, Hume is describing what he takes to be an “opposition” within the imaginationw, between the understanding, which gives rise to conclusions from cause and effect, and the imaginationn. Hume appeals to this opposition at pages 265–66: The memory, senses, and understanding are, therefore, all of them founded on the imagination, or the vivacity of our ideas. No wonder a principle so inconstant and fallacious shou’d lead us into errors, when implicitely follow’d (as it must be) in all its variations. ’Tis this principle, which makes us reason from causes and effects; and ’tis the same principle, which convinces us of the continu’d existence of external objects, when absent from the senses. But tho’ these two operations be equally natural and necessary in the human mind, yet in some circumstances they are1 directly contrary. (265–66) Hume’s footnote is to I.iv.4. The effects of holding such opposing or contradictory beliefs are illuminated earlier in Part IV of Book I: “any contradiction . . . gives a sensible uneasiness,” from which “the mind . . . will naturally seek relief” (T 205–6). Hume characterizes such “contradiction” as an “opposition” or “combat” (205–6). Similarly, he describes a “contradiction” that involves “struggle and opposition” between “two enemies,” a struggle from which we “endeavour to set ourselves at ease” (215). In these passages, Hume intends the language of psychological conflict literally. Hume maintains that when there is an “opposition of two contrary principles,” the mind “must look for relief by sacrificing one to the other” (206). In other words, there will be an inclination to remove the contradiction by relinquishing one of the opposing beliefs. Contradictions may be logical, but they give rise to an inclination to revise one’s beliefs, so that a belief system containing the contradiction is in disequilibrium. In the case of the opposition introduced at I.iv.4, the disequilibrium cannot be removed. The passage at pages 265–66 continues: How then shall we adjust those principles together? Which of them shall we prefer? Or in case we prefer neither of them but successively assent to both, as is usual among philosophers, with what confidence can we afterwards usurp that glorious title, when we thus knowingly embrace a manifest contradiction? (T 266) It is natural to think of the successive assent to members of a pair of contradictory beliefs as resulting from concurrent, but conflicting, dispositions to assent to each member of the pair (215). Hume often writes of a “propensity”

190

reflection and the stability of belief

or “propension” (199, 209, 210, 218, 253, 255), or “inclination” (154, 208, 210, 238; cf. 224), to hold a belief. In cases where these dispositions conflict, the mind is “divided within itself” (154) or “divided betwixt . . . opposite principles” (238). Philosophers would not “successively assent to both” members of a pair of contradictory beliefs if either of the dispositions to assent were avoidable or resistible. Where both dispositions are unavoidable and irresistible, the inclination to relinquish one of the opposing beliefs is frustrated. The result is an inescapable disequilibrium, manifested as alternating assent.7 Hume’s claim at page 266 is that the imaginationw generates ineliminable instability. This initial consideration does not complete Hume’s case for rejecting all belief as unjustified. Hume writes: “This contradiction wou’d be more excusable, were it compensated by any degree of solidity and satisfaction in the other parts of our reasoning” (T 266). His point, I think, is not that the instability he claims to have located at pages 265–66 is excusable, but rather that its negative import for justification could in principle be mitigated. This instability, after all, is localized in that it arises in connection with a single, albeit fundamental, proposition—that bodies have a continued and independent existence. This leads Hume to introduce some additional considerations at pages 266–68, before announcing his readiness to reject all belief as unjustified. Much of what he says here also suggests a crucial role for equilibrium. I have in mind features of the “very dangerous dilemma” discussed at pages 267–68. The dilemma draws on material at I.iv.1, “Of skepticism with regard to reason.” At the first three paragraphs of this section, Hume argues that “all knowledge,” even demonstrative knowledge, “degenerates into probability” (T 180). At paragraphs five through seven, Hume argues that judgments of probability are subject to correction in light of the fallibility of judgment. Hume thinks this correction takes the form of a reduction in the estimate of probability. Such a reduction results in a new judgment of probability that is itself subject to correction and reduction, and so on ad infinitum. The upshot is that judgments of probability are subject to a series of corrections and reductions, which, if fully sustained, would result in “a total extinction of belief and evidence” (183). Judgments of probability, when subjected to a sustained series of corrections, degenerate into the absence of any belief at all. Hume repeats this conclusion: “[T]hese . . . principles [of judgment] when carry’d farther, and apply’d to every new reflex judgment, must, by continually diminishing the original evidence, at last reduce it to nothing, and utterly subvert all belief and opinion” (184). “Of scepticism with regard to reason” concerns “reason” or the “understanding” in the sense of the notes at pages 117–18 and 371—demonstrative and probable reasonings. It is the sustained operation of the “understanding” (182, 187) that subverts itself. 7. This result can be mitigated if the mind finds a way to obscure the contradiction (see my discussion later in §2).

hume on stability, justification, and unphilosophical probability

191

It is against this background that Hume presents “a very dangerous dilemma, which-ever way we answer it” (T 267): either we assent to the imaginationn as well as to the understanding, or we adhere solely to the understanding. Equilibrium cannot be achieved in either case. Here is what Hume says about the first horn of the dilemma: For if we assent to every trivial suggestion of the fancy; beside that these suggestions are often contrary to each other; they lead us into such errors, absurdities, and obscurities, that we must at last become asham’d of our credulity. Nothing is more dangerous to reason than the flights of the imagination, and nothing has been the occasion of more mistakes among philosophers. . . . This has already appear’d in so many instances, that we may spare ourselves the trouble of enlarging upon it any farther. (267) Suppose we accept the trivial suggestions of the fancy, that is, the imaginationn. These are “often contrary to each other,” and hence produce disequilibrium. They also “lead us into . . . errors, absurdities, and obscurities.” Our inclination to relinquish some member or a pair of opposing beliefs often meets with resistance—there is a “reluctance” (206) to give up the beliefs, or the beliefs prove “obstinate” (215). This is true even if one of the opposing beliefs is resistible or avoidable. In the face of resistance, we seek to revise our doxastic system by adding a new belief that functions to obscure the contradiction. These obscurities are by-products of the disequilibrium.8 In writing at page 267 of “absurdities, and obscurities,” Hume would have had in view his discussion (two sections earlier) of whether taste admits of local conjunction with matter: Here then we are influenc’d by two principles directly contrary to each other, viz. that inclination of our fancy by which we are determin’d to incorporate the taste with the extended object, and our reason, which shows us the impossibility of such an union. Being divided betwixt these opposite principles, we renounce neither one nor the other, but involve the subject in such confusion and obscurity, that we no longer perceive the opposition. We suppose, that the taste exists within the circumference of the body, but in such a manner, that it fills the whole without extension. (238) In effect, we believe “that a thing is in a certain place, and yet is not there” (238). This belief obscures the underlying conflict, without resolving it, so that the belief is an “absurdity” (238). Assenting to the imaginationn as well as to the understanding results in disequilibrium.

8. For additional discussion, see my 1991, esp. §4.

192

reflection and the stability of belief

Suppose we adhere to the understanding alone. This leads to the second horn of the dilemma: But on the other hand, if the consideration of these instances makes us take a resolution to reject all the trivial suggestions of the fancy, and adhere to the understanding, that is to the general and more establish’d properties of the imagination; even this resolution, if steadily executed, wou’d be dangerous, and attended with the most fatal consequences. For I have already shewn,1 that the understanding, when it acts alone, and according to its most general principles, entirely subverts itself, and leaves not the lowest degree of evidence in any proposition, either in philosophy or common life. (T 267–68) Hume observes that he had already shown (his footnote is to I.iv.1) that the sustained operation of the understanding would entirely subvert itself, with the result that all judgments of probability perish. Adhering to the understanding alone cannot achieve equilibrium in belief, except in the degenerate sense of destroying all belief. Let me summarize the import of I.iv.7 for my interpretation. Hume’s initial consideration, en route to declaring his readiness to reject all belief as unjustified, is that the imaginationw generates contradictory beliefs about the existence of body that are unavoidable and irresistible, and hence leads to inescapable disequilibrium. Hume does not find anything to mitigate this obstacle to justified belief. To the contrary, instabilities that systemically infect the understanding and imaginationn are in the offing. This is the point of the dilemma. The understanding acting alone would subvert itself, destroying all belief. Assenting to the imaginationn as well as to the understanding results in belief, but at the expense of equilibrium. One cannot have equilibrium in either case—because either one lacks belief, or one’s beliefs are in disequilibrium. Unlike the initial consideration, this is not a localized difficulty. The dilemma is presented as a general result about the faculties of the imaginationn and the understanding. Much of what Hume says at pages 265–68 supports the idea that Hume’s readiness to reject all belief as unjustified should be explained on the hypothesis that he explicates justification with reference to equilibrium.9 This provides a prima facie case for my interpretive framework. At I.iv.7, Hume claims to have discovered that his theory of justification fails to sustain his pretheoretical epistemological commitments. Hume’s response at pages 268–74 to this destructive result is complex, but at no point does he disclaim, even implicitly, the importance of facts about instability in evaluating these commitments. He

9. Hume does consider, and reject, a possible resolution of the dilemma. I have in mind the passage at page 268 beginning “We save ourselves” and continuing through the end of the paragraph. I believe Hume’s reasons for rejecting the solution can themselves be reconstructed with reference to equilibrium.

hume on stability, justification, and unphilosophical probability

193

retains his picture of justification and avows that our considered epistemological judgments cannot be sustained. Before turning to the topic of unphilosophical probability, I note that the developments at pages 265–68 pose a serious difficulty for the prominent line of interpretation deriving from the work of Norman Kemp Smith. Kemp Smith maintains that Hume has a response to skepticism in his doctrine of “natural beliefs” that are “inevitable” or “irresistible.”10 This leads Kemp Smith to formulate a sense in which, for Hume, natural beliefs ought to be accepted: “The beliefs which ought to be accepted are, [Hume] teaches, beliefs that Nature itself marks out for us. In their fundamental forms, as ‘natural’ beliefs, we have no choice but to accept them; they impose themselves upon the mind.”11 John Lenz, P. F. Strawson, and Barry Stroud have adopted versions of Kemp Smith’s view.12 The common thread among these commentators can be put as follows: to say that we ought not hold the natural beliefs is at best pointless, since we irresistibly or unavoidably do hold them; and at worst false, if the claim that we ought not hold them implies that we are able not to hold them. I call this the Kemp Smith interpretation. At pages 265–66, in introducing the “initial consideration” discussed previously, Hume brings himself to conclude that our belief-forming mechanisms generate contradictory beliefs that are unavoidable and irresistible. After observing that the resulting instability is manifested as alternating assent, Hume is poised to announce his readiness to reject all belief as unjustified. The Kemp Smith interpretation cannot explain how irresistible beliefs can fail to be justified; in the case of any irresistible belief, it is at best pointless to say that we ought not hold it. But Hume does consider irresistible beliefs unjustified when they are parties to disequilibrium (or, perhaps, when they result from mechanisms that tend to produce disequilibrium), even though they are irresistible. On Hume’s theory of justification, unlike the Kemp Smith account of it, irresistibility is not a sufficient condition for justification.13

10. Kemp Smith 1941, 87, 455, 486, and 1905, esp. 152, 161, 162. 11. Kemp Smith 1941, 388 (cf. 46, 68), and 1905, 152. 12. See Lenz 1958, 559, 566–67; P. F. Strawson 1958, 20–21, and 1985, 10, 11; and Stroud 1977, 76, 247 (cf. 248). 13. Kemp Smith was not oblivious to the difficulty. His response is that “this is a more sceptical conclusion than is strictly demanded by Hume’s philosophy” (1905, 168; cf. 1941, 128). What Hume ought to have said, according to Kemp Smith’s pioneering article, is that the relevant belief-forming mechanisms have “limited range” (1905, 169). In his subsequent book, he portrays Hume as himself drawing the conclusion that the relevant mechanisms “are reliable and legitimate only within a strictly limited domain,” and that there is “an interdict against the universalising of either of them” (1941, 128, 493–94). This misrepresents Hume’s position. Kemp Smith’s discussion refers to Treatise “265 ff.” (1941, 493n.2). It is at page 266 that Hume maintains that we can do no better than successive assent to the contradictory principles. This differs from the suggestion that the belief-forming mechanisms apply within a limited domain. Furthermore, Kemp Smith provides no explanation of how the unavoidable and irresistible belief-forming mechanisms that generate the contradiction are interdicted or contained.

194

reflection and the stability of belief

3. The Fourth Species of Unphilosophical Probability—Instabilities due to Observation of Accidental Conjunctions I turn to the bearing of Hume’s theory of justification on his discussion of unphilosophical probability. Hume writes at the beginning of I.iii.13: “All these kinds of probability are receiv’d by philosophers, and allow’d to be reasonable foundations of belief and opinion. But there are others, that are deriv’d from the same principles, tho’ they have not had the good fortune to obtain the same sanction” (T 143). The first sentence refers to the kinds of probability discussed in the preceding two sections—the probability of chances, of causes, and arguments from analogy; we can say they constitute “philosophical probability.” The kinds of probability that are not “receiv’d by philosophers” are the topic of I.iii.13. Hume proceeds to enumerate four species of unphilosophical probability. One species faces “the opposition of philosophy” (143); another is “disclaimed by philosophers” (143). At page 143, Hume does not explicitly associate himself with the distinction between philosophical and unphilosophical probability; he attributes the distinction to “philosophy” and “philosophers.” Hume assumes the same posture toward other epistemological distinctions. Education is “never . . . recogniz’d by philosophers” (T 117); it is “disclaim’d by philosophy, as a fallacious ground of assent to any opinion” (118). Hume distinguishes between “principles which are permanent, irresistable, and universal” and “the principles, which are changeable, weak and irregular,” observing that “the former are received by philosophy, and the latter rejected” (225). In claiming that philosophy rejects education, unphilosophical probability, and principles that are changeable, weak, and irregular, Hume does not name the philosophers he has in mind. They are presumably the natural philosophers.14 Since Hume seeks to “introduce the experimental method of reasoning into moral subjects” (xi), we must suppose that he is himself inclined to accept these distinctions. At the same time, in attributing them to “philosophy” and unnamed “philosophers,” Hume distances himself from these commitments, holding them at arm’s length. This device reflects Hume’s ambivalence toward these distinctions and enables him to discuss the distinctions without either owning or disowning them. My interpretation explains this ambivalence, much as it explains Hume’s contradictory assessments (as discussed in §2) of a number of epistemological distinctions. In this sense, the contradictory and ambivalent assessments are of a piece. The constructive stage of the development of the theory of justification predominates when Hume explicitly associates himself with the claim that beliefs based on causal inference are justified (as at T 89, 107–8, 225–26). The destructive stage of the development of the theory prevails 14. I owe this suggestion to Jan Ludwig.

hume on stability, justification, and unphilosophical probability

195

when Hume announces his readiness to reject all belief as unjustified (268–69). Both stages are in view when Hume registers ambivalence in regard to epistemological distinctions (as at 117–18, 143, 225). On the one hand, Hume thinks he achieves some provisional success in explaining his pretheoretical commitments. On the other hand, he knows he will reject these distinctions when he reaches the conclusion that no belief is justified. Although Hume embraces this pessimistic evaluation of the program of sustaining his pretheoretical distinctions, the program is his own. Hume must think there is a presumptive basis for the distinction between philosophical and unphilosophical probability. My objective in §§3–4 is to understand the basis for this presumption. Hume’s reason, at I.iii.13, for rejecting unphilosophical probability as unreasonable is not that all belief is rejected as unreasonable—Hume’s pessimistic conclusion that all belief is unjustified lies down the road, at I.iv.7. His reason is that unphilosophical probability is unreasonable in comparison to philosophical probability. In the penultimate paragraph of I.iii.13, Hume reviews the kinds of probabilities he has discussed: proofs, the probability of causes, and the four species of unphilosophical probability. In each case, “’Tis by habit we make the transition from cause to effect” (T 153–54), though probabilities, as distinct from proofs, result from a habit that is weak or imperfect (130–31, 133, 142, 153–54). Philosophical and unphilosophical probability are “deriv’d from the same principles,” custom or the relation of cause and effect. On what basis are some probabilities based on causal inference rejected as unreasonable at this stage of the Treatise? My interpretation suggests an explanation. Instabilities can be eliminated in some cases, but not in others, and this difference matters from the perspective of the theory of justification. Although Hume will eventually contend (at I.iv.7) that causal inference and the imaginationn give rise to inescapable disequilibrium, at I.iii.13 he takes himself to have located disequilibria within causal inference that are eliminable. This discovery of eliminable disequilibria does not lead Hume to reject all belief based on causal inference as unjustified. In this section, I consider the application of this explanation to the fourth kind of unphilosophical probability; I discuss the first three kinds in the next section. Hume discusses the fourth kind of unphilosophical probability at pages 146–50.15 He begins with an example: “A fourth unphilosophical species of probability is that deriv’d from general rules, which we rashly form to ourselves, and which are the source of what we properly call Prejudice. An Irishman cannot have wit, and a Frenchman cannot have solidity” (T 146). As at pages 153–54, Hume attributes the “rash” general rules to “those very principles, on which all judgments concerning causes and effects depend,” to 15. There is little discussion of the fourth kind of unphilosophical probability. The exceptions include Fogelin 1985, 60–63; Hearn 1970, esp. 405–14; Hendel 1925, 204 (or 1963, 170–71); MacNabb 1951/1966, 97–99; and Passmore 1952/1968, 54–64, passim, and 1977, esp. 83–84.

196

reflection and the stability of belief

“habit and experience” (147). “Rash” cannot simply mean based on too small a sample. Hume allows that generalizations based on a single experience can be reasonable (104–5, 131, 175).16 Then why is the probability derived from general rules “unphilosophical”? The following passages suggest an answer: [T]ho’ custom be the foundation of all our judgments, yet sometimes it has an effect on the imagination in opposition to the judgment, and produces a contrariety in our sentiments concerning the same object. (T 147–48) We shall afterwards take notice of some general rules, by which we ought to regulate our judgment concerning causes and effects. . . . By them we learn to distinguish the accidental circumstances from the efficacious causes; and when we find that an effect can be produc’d without the concurrence of any particular circumstance, we conclude that that circumstance makes not a part of the efficacious cause, however frequently conjoin’d with it. But as this frequent conjunction necessarily makes it have some effect on the imagination, in spite of the opposite conclusion from general rules, the opposition of these two principles produces a contrariety in our thoughts, and causes us to ascribe the one inference to our judgment, and the other to our imagination. The general rule is attributed to our judgment; as being more extensive and constant. The exception to the imagination; as being more capricious and uncertain. Thus our general rules are in a manner set in opposition to each other. When an object appears, that resembles any cause in very considerable circumstances, the imagination naturally carries us to a lively conception of the usual effect, tho’ the object be different in the most material and most efficacious circumstances from that cause. Here is the first influence of general rules. But when we take a review of this act of the mind, and compare it with the more general and authentic operations of the understanding, we find it to be of an irregular nature, and destructive of all the most establish’d principles of reasoning; which is the cause of our rejecting it. This is a second influence of general rules, and implies the condemnation of the former. . . . [T]he sceptics may here have the pleasure of observing a new and signal contradiction in our reason, and of seeing all philosophy ready to be subverted by a principle of human nature, and again sav’d by a new direction of the very same principle. (149–50) As we have seen (§2), elsewhere in the Treatise Hume describes contradictions as involving “opposition” or “combat” (205–6), and “struggle and 16. This tells against Fogelin’s account of “rash” generalizations as based on too small a sample (1985, 61).

hume on stability, justification, and unphilosophical probability

197

opposition” between “two enemies” (215). With these passages in view, I take the language of psychological conflict at pages 147–48 and 149–50— “opposition” and “contrariety,” principles that are “destructive,” and philosophy being “subverted”—as evidence that considerations about equilibrium are central to Hume’s treatment of the fourth kind of unphilosophical probability. The substance of Hume’s account is nevertheless unclear; the roles of accidental circumstances, and of the two “influences” of general rules, require explanation. At page 148, after remarking for the first time that custom can produce contrariety, Hume calls attention to the observation of accidental conjunctions. It is helpful to formulate the explanation schematically. Suppose there is an observed constant conjunction both between the circumstances A and C, and between A and C and some subsequent effect E; that C is the “essential” or genuine cause of E; and that it is “by accident” that A is constantly conjoined with C, and hence with E—the presence of A is “superfluous” to the production of E. In saying that C, but not A, is the genuine cause of E, Hume is offering an intuitive judgment; he is not describing the implications of a theory of causation. Hume points out that upon the observation of a new instance of A, even in the absence of an instance of C, custom leads to the belief that an instance of E will occur. This belief may be mistaken, but Hume has not yet exhibited any “contrariety” in beliefs. What is required is the supposition that the new instance of A is also accompanied by an observation of an instance of some circumstance D, such that there is an observed constant conjunction between D and the nonoccurrence of E. Under these conditions, we will have the inclination to believe both that an instance of E will occur and that an instance of E will not occur. Hume’s example at page 148 of a man in a suspended tower can be adapted to this schema. Suppose the man has observed that whenever he sees a precipice and has not been suspended, he has fallen. His not being suspended (C) was the genuine cause of his falling (E), and his seeing the precipice (A) superfluous; it is by accident that in his experience seeing a precipice is constantly conjoined with his falling. Suppose the man has observed that whenever he is suspended (D), he has not fallen (not-E). Suppose that for the first time the man both sees a precipice (A) and is suspended (D). He will have the inclination to believe both that he will fall and that he will not fall, so that there is “a contrariety in [his] sentiments concerning the same object.” This conflict has its source in two observed conjunctions—(1a) “whenever I see a precipice (A), I fall (E),” and (1b) “whenever I am suspended (D), I do not fall (not-E).” Although (1a), unlike (1b), is accidental, both generalizations result from custom and contribute equally to the disequilibrium. The problem is to find a principled way of explaining why generalizations conditioned by observation of accidental conjunctions are unjustified.17 17. It is of no help that Hume’s psychological theory allows that belief is more strongly conditioned by a greater number of instances. There is nothing to preclude the possibility of observing equal numbers of instances of two constant conjunctions, one accidental and one not.

198

reflection and the stability of belief

Hume’s solution is to maintain that the method for removing the instability favors the nonaccidental generalizations. It is here that the discussion of the two “influences” of general rules comes into play. It is tempting to think that at pages 149–50 “the first influence of general rules” and “the second influence of general rules” are nothing more than labels for those generalizations that are accidental and those that are not, without any rationale for relying on one rather than the other. I believe, however, that Hume is appealing to different levels or orders of beliefs about observed conjunctions. (The “second influence” of general rules arises “when we take a review of [the] act of the mind” that constitutes the “first influence of general rules.” Similarly, at page 148, there is a role for “reflection” on the “circumstances” in which we have a propensity to rely on accidental generalizations.) First-order generalizations are extrapolations from observed conjunctions between the members of resembling pairs of objects. These generalizations are sometimes falsified by subsequent observation. Second-order generalizations are extrapolations from the observed success of classes of first-order generalizations in avoiding falsification.18 The “first” and “second influence” of general rules derive from beliefs based on first- and second-order generalizations, respectively. In order to see the relevance of the two “influences,” it is helpful to consider that, for Hume, generalizations are habits of expectation. Generalizations of different orders are habits of expectation of different orders. The strength of a habit of expectation is, ceteris paribus, proportional to the number of observed instances on which the habit is based. The strength of a habit can be affected by higher-order habits. For example, a second-order habit of expecting the chemical composition of an ingested substance to be correlated with changes in health tends to reinforce or strengthen a firstorder habit of expecting an arsenic ingestion to be followed by death. Or suppose some poison has a distinctive color. A second-order habit of expecting that the color of an ingested substance is not correlated with changes in health conflicts with a first-order habit of expecting an ingestion of a substance of the distinctive color to be followed by death. (This is the sense in which “the first influence of general rules” can “be of an irregular nature, and destructive of all the most establish’d principles of reasonings.” They can conflict with second-order generalizations that are “more extensive and constant” in that they are about classes of first-order generalizations.) As in the case of conflicts between two first-order generalizations, first- and second-order generalizations implicated in such conflict arise from custom.19 In these cases, the second-order habit tends to weaken the first-order habit. 18. Here I disagree with MacNabb’s claim that “these general rules [of the second sort] are . . . not generalizations, as far as Hume here suggests, about the successfulness or the reverse of different kinds of expectations” (1951/1966, 98). 19. Hearn is not correct that the contrast between the first and second kinds of general rules is that the latter are essentially reflective in the sense of being “consciously formulated and adopted” (1970, 410–11). Neither kind of general rule need be reflective, and either can be reflective. Similarly, contrary to Hearn, the second kind of general rule is not essentially “directive” (411) or regulative.

hume on stability, justification, and unphilosophical probability

199

This weakening of the first-order habit mitigates the conflict, so that it is only “in a manner” that first- and second-order general rules are “set in opposition to each other” (T 149). Conflicts between first-order generalizations can be resolved because first-order habits tend to be strengthened, or weakened, in the presence of relevant second-order habits. This is the “second influence” of general rules, that is, the effect of second-order generalizations. In the case of the suspended man, the conflict between (1a) and (1b) can be resolved by locating a second-order generalization that will either weaken or strengthen one of the first-order generalizations. Suppose, for example, the suspended man believes the second-order generalization (2): “first-order universal generalizations relating resemblances in height (without reference to systems of suspension) to resemblances in fall or descent are falsified.” (The man might believe this on the basis of his experiences with modest heights, with the fall of objects other than his own body, et cetera) He will then experience two conflicts—both between his inclinations to believe (1a) and (1b) and between his inclinations to believe (1a) and (2). In the presence of (2), however, both conflicts will be resolved (or at least mitigated), for (2) will weaken the inclination to believe the accidental first-order generalization, (1a), which is a party to both conflicts.20 There is an asymmetry between (1a) and (1b), in that equilibrium is restored when (1a) is weakened. The cases of prejudice also involve the observation of accidental conjunctions. The first-order generalization “all Irishmen lack wit” can be weakened by the second-order generalization “universal generalizations relating resemblances in national origin to resemblances in intellectual characteristics are falsified.” Accidental generalizations give rise to conflicts with nonaccidental generalizations, and hence to disequilibrium. Hume maintains that the accidental generalizations will be weakened, and the nonaccidental generalizations strengthened, by higher-order generalizations, so that the conflicts to which accidental generalizations give rise are resolved in the presence of relevant second-order generalizations.21 Acceptance of accidental, as opposed to nonaccidental, generalizations is “unphilosophical” because it leads to eliminable

20. It might be objected that Hume treats the “influence” of second-order generalizations as varying among individuals: “This is a second influence of general rules, and implies the condemnation of the former. Sometimes the one, sometimes the other prevails, according to the disposition and character of the person. The vulgar are commonly guided by the first, and wise men by the second” (T 150). I believe Hume’s point is that “passion” or the “affections” can, by enlivening the idea derived from the accidental generalization, overwhelm the influence of second-order generalizations (148). Thus, the suspended man’s fear of falling enlivens his idea of descent (148). When Hume claims that we “rashly” (146) form the accidental generalizations, I think he means that we do so in accord with our passions, thereby distorting the influence of second-order generalizations. 21. Quine relies on second-order inductions for a similar purpose, to revise our innate standards of similarity or natural kinds in order to achieve scientifically more sophisticated groupings. See Quine 1969, esp. 127–29.

200

reflection and the stability of belief

instability.22 This account coheres with much that Hume says elsewhere. He explicitly appeals to a higher-order habit to explain how the experience of a single instance of a conjunction can produce beliefs about unobserved instances (T 105). Hume considers reliance (under appropriate conditions) on this higher-order habit justified. The higher-order habit enables us to “attain the knowledge of a particular cause” (104), and it is “the source of most of our philosophical reasonings” (173). Hume also appeals to higherorder habits to explain how we form judgments about minute differences in probability. In this context, the higher-order habits are explicitly associated with “general rules” (141–42). At page 142, Hume notes that “these general rules we shall explain presently,” that is, at pages 146–50. This completes my reconstruction of Hume’s treatment of the fourth kind of unphilosophical probability. The question arises as to why so much reconstruction is required. The answer relates to the tensions in Hume’s attitude toward the epistemological distinctions he draws. Hume is more concerned, at pages 146–50, to foreshadow the destructive stage of his development of his theory of justification than to explain in detail how the instabilities associated with the fourth kind of unphilosophical probability can be removed. We have seen that the notions of “contrariety,” “opposition,” and “subversion” play a role in Hume’s description of the fourth kind of unphilosophical probability at pages 149–50. These notions will also be prominent in Hume’s discussion of instabilities at pages 265–68 of I.iv.7, and in the sources of these discussions at I.iv.1 and I.iv.4 (T 184, 231). At I.iv.7, when Hume will be ready to reject all belief as unjustified, it is because he takes himself to locate inescapable instabilities. The fourth kind of unphilosophical probability is a locus of eliminable instability; philosophy is “ready to be subverted,” but can be “sav’d”—the second influence of general rules can contain the “destructive” influence of the first. It is thus “the sceptics,” rather than Hume, who are said to “have the pleasure of observing a new and signal contradiction in our reason” (150). It is Hume, however, who chooses to seize the discovery of an eliminable instability as a harbinger of the ineliminable instabilities, the “manifold contradictions and imperfections” (268), that he will bring forward in I.iv.7. Although Hume’s epistemological commitment to causal inference is evident as late as pages 225–26, as early as page 150 he is preparing the ground for the pessimistic conclusion that will emerge at I.iv.7.23 22. What is to rule out the existence of an accidental higher-order generalization that strengthens an accidental generalization, or weakens a nonaccidental generalization, of a lower order? Hume might contend this is improbable. Or we should perhaps locate Hume in the (Ramsey-Lewis) tradition on which nonaccidental regularities are identified with those that occupy a place in a suitably ideal theoretical system of generalizations. (See Lewis 1973, 72–77.) It is a stock objection to Hume that, in offering a regularity analysis of causation, he fails to consider counterexamples deriving from accidental conjunctions. Hume exhibits sensitivity to the possibility of accidental regularities at pages 4, 104–5, 146–50, and 175 of the Treatise. The discussion of the fourth kind of unphilosophical probability is the place to look for resources for a Humean reply. 23. Fogelin offers similar observations (1985, 60–62).

hume on stability, justification, and unphilosophical probability

201

4. The First Three Species of Unphilosophical Probability—Instabilities due to “Psychological Distance” I turn to an inventory of the first three kinds of unphilosophical probability, discussed by Hume at pages 143–46.24 (i) The vivacity of perceptions of memory diminishes over time. Since belief always attends or accompanies memory (T 86), the vivacity of memory beliefs diminishes over time. Hume takes this to imply that we have a greater degree of “assurance” (143, 144) in the memory of a recent experience than in the memory of a more remote experience. Thus, doxastic states produced by memory are unstable in that they have a tendency to change gradually with respect to the degree of assurance with which they are held. This tendency to change is entirely the result of the associationist mechanism that sustains the memories. (ii) In causal inference, vivacity is communicated from a lively perception to an associated idea (T 144, 145). We have from (i) that the vivacity of memory diminishes over time. A causal inference based on memory of a recent experience therefore results in a more vivacious belief than a causal inference based on memory of a remote experience.25 Hume takes this to imply that a causal inference founded upon a matter of fact we remember is more “convincing” (143) in the former case than in the latter.26 This second kind of unphilosophical probability is parasitic on the first. The difference in the degree of vivacity of belief resulting from causal inferences founded on memories of events differing in remoteness is due entirely to the difference in the degree of vivacity of the memories themselves. For this reason, I have reversed the order of Hume’s discussion; what is for Hume the “first” (“second”) kind of unphilosophical probability is, in my discussion, the “second” (“first”). In the penultimate paragraph of I.iii.13, Hume describes unphilosophical probability as arising from habit, or custom. This is not quite accurate. Unphilosophical probability arises from memory or habit; alternatively, it arises within the first and second systems of beliefs described by Hume at pages 109–10 (§2). Habit is the source of the third kind of unphilosophical probability. (iii) In causal inference, a present perception communicates some, but not all, of its vivacity to the associated idea (T 98, 144). In a chain 24. There is virtually no discussion of the first three kinds of unphilosophical probability. The few exceptions are confined to one or two pages: Fogelin 1985, 60; Hendel 1925, 203–4 (or 1963, 170–71); Laird 1932, 91; MacNabb 1951/1966, 96–97; and Passmore 1952/1968, 59–60. 25. There is a suppressed assumption about the proportion of the vivacity of a present perception that is communicated to the associated idea: that it does not vary inversely with the vivacity of the present impression. 26. Hume fails to note a generalization of this idea. Memory is less vivacious than senseperception (T 8). It follows that a causal inference based on memory of an event results in a less vivacious belief than the same causal inference based on sense-perception of the same event at the time the event occurs.

202

reflection and the stability of belief

of connected causal inferences, beliefs in conclusions at later points in the chain are less vivacious than the beliefs in conclusions at earlier points. Hume takes this to imply that “conviction” or “persuasion” is greater the shorter the chain (144). Hume is calling attention to what he takes to be empirical phenomena that can be described nonassociationistically: other things being equal, long chains of causal inferences produce less assurance than short chains. In the two cases involving memory, degree of assurance is inversely proportional to the remoteness of the event, to temporal distance. In the case of the third kind of unphilosophical probability, where “the imagination is carry’d thro’ a long chain of connected arguments,” the vivacity of the original perception “must gradually decay in proportion to the distance” (T 144). “Distance” must refer to the number of inferences in the chain, to the length of the argument; degree of assurance is inversely proportional to “argumentative distance.” Degree of assurance diminishes as what I call psychological distance increases, where argumentative length and remoteness of memory are two sources of psychological distance. Hume’s associationist principles governing the degradation of vivacity are intended to explain this general empirical phenomenon.27 Hume owes us, in addition to his associationist account of why the variations in degree of assurance occur, an account of the presumptive basis for considering them unphilosophical. Hume comments, with reference to the second kind of unphilosophical probability: “the difference in these degrees of evidence [is] not receiv’d by philosophy as solid and legitimate; because in that case an argument must have a different force to-day, from what it shall have a month hence” (T 143). The argument that fire accompanied a particular instance of smoke will generate less assurance ten months after the fire than ten weeks after the fire. Hume writes, with reference to the third kind of unphilosophical probability: “a man may receive a more lively conviction from a probable reasoning, which is close and immediate, than from a long chain of consequences, tho’ just and conclusive in each part” (144). Consider a large number of inferential transitions, such that each transition on its own is “infallible” (144); Hume has in mind causal inferences that constitute “proofs” (§2). If such inferences are constructed into a long chain of connected arguments, one will have relatively little assurance in the conclusion. Since the degree of assurance in a conclusion is a function of the degree of assurance in each step of an argument, an inference that is infallible 27. Hume’s associationist interests help explain why he presents second a kind of unphilosophical probability that is parasitic on the one he presents first. The reason is tactical. In earlier sections of the Treatise, Hume has extensively discussed the kinds of associationist principles he requires, in their application to the association of ideas by the relation of cause and effect. Earlier sections contain much less foundation for treating memory as an associative process, governed by similar principles. Although the complete associationist explanation of the kind of unphilosophical probability presented first presupposes the explanation of the kind presented second, discussion of the former, where association by causation has a role, prepares the reader for the associationist explanation of the latter.

hume on stability, justification, and unphilosophical probability

203

on its own is not infallible when connected in a chain. The degree of assurance generated by a causal inference embedded in an argument varies with the number of preceding steps. As D. G. C. MacNabb notes, Hume’s general claim is that, as a result of the variations in vivacity, the same argument has a different force, produces different degrees of assurance, at different times or on different occasions of use. MacNabb suggests that the first three kinds of unphilosophical probability are to be rejected for this reason.28 This is the most promising suggestion that is explicit at pages 143–46. As the suggestion stands, however, it will not take Hume far. The claim that, in the described circumstances, the same argument has a different force on different occasions is questionbegging. What is to count as “the same argument”? Hume’s claim is correct if we think of an argument as consisting simply of a set of propositions comprising premises and a conclusion. Hume’s claim is incorrect if we think of an argument as including a set of premises together with the degrees of assurance with which the premises are believed. The first three kinds of unphilosophical probability arise in circumstances where there are variations in this assurance. Why should arguments not have a different force on different occasions, if they differ in the degree of assurance associated with the beliefs in the premises? My own account of why the first three kinds of unphilosophical probability are unjustified is modeled on aspects of Hume’s treatment of poetical enthusiasm and of the relations of resemblance and contiguity. (When I write of the relation of resemblance, the discussion applies to contiguity as well.) Hume cannot raise the issue of whether beliefs arising from resemblance and poetical enthusiasm are unreasonable. Resemblance can enliven an idea, or intensify a belief, but not produce belief (T 107–10); poetical enthusiasm enlivens ideas (123, 630–32/1.3.10.10–12) but results merely in “phantom” (630/1.3.10.10) or “counterfeit belief” (123).29 I shall say that resemblance and poetical enthusiasm produce quasi-belief. At the same time, the question of the extent to which a belief-forming mechanism produces justified beliefs is a special case of the question of the extent to which a vivacity-producing mechanism produces epistemically appropriate degrees of vivacity. Hume can raise this latter question with respect to resemblance and poetical enthusiasm. I begin with a prior question: how does the quasi-belief that results from resemblance and poetical enthusiasm differ from belief? My working hypothesis is that Hume has a single answer for the two cases. The answer cannot be that beliefs are more vivacious than quasi-beliefs. Hume writes of the effects of poetry: [H]ow great soever the pitch may be, to which this vivacity rises, . . . ’tis still the mere phantom of belief or persuasion. . . . A poetical 28. See MacNabb 1951/1966, 96–97. 29. [For a textual note on Treatise 123, see “References to Hume,” this volume.]

204

reflection and the stability of belief

description may have a more sensible effect on the fancy, than an historical narration. . . . It may seem to set the object before us in more lively colours. (T 630–31/1.3.10.10) We can suppose that an occurrent belief must meet or exceed some threshold of vivacity, that it must be “vivacious.” Ideas arising from poetical enthusiasm can meet such a threshold. In order to distinguish quasi-belief from belief, it is necessary to rely on Hume’s well-known tendency to treat belief as a disposition, as well as an occurrence.30 A disposition to vivacity is a disposition to experience vivacious ideas, ideas that possess the degree of vivacity required for occurrent belief.31 Some dispositions to vivacity are unstable in that they have a tendency to change abruptly, simply because they are not produced by perception, memory, or repetition. Such dispositions, in Hume’s terminology, lack fixity. Hume in effect stipulates that a dispositional belief is an infixed disposition to vivacity; a dispositional quasibelief is a disposition to vivacity that is not infixed.32 A vivacious occurrent idea is an occurrent belief if it manifests a disposition that constitutes a belief; it is an occurrent quasi-belief if it manifests a disposition that constitutes a quasi-belief. Vivacity arising from resemblance or poetical enthusiasm, no matter how intense, does not constitute occurrent belief, because it does not arise from a disposition that is infixed. Let me confirm this interpretation. The association of ideas by resemblance produces “momentary glimpses of light” (T 110). The momentary or fleeting character of these ideas is a symptom of quasi-belief but not its distinguishing feature. A poetical enthusiasm can be sustained. What matters is that “the least reflection dissipates the illusions of poetry” (123). This is an observation about the stability of the underlying disposition. Quasi-beliefs can be relatively permanent, but they tend to be fleeting, and hence tend to change abruptly. This is Hume’s point when he describes belief as “firm” (97, 105, 106, 116, 624, 626, 629/1.3.7.7, 631/1.3.10.10) and “solid” (106, 110, 121, 629/1.3.7.7, 631/1.3.10.10). Hume is also getting at the importance of the stability of the underlying disposition to vivacity in his explanations of why resemblance and poetical enthusiasm do not produce belief. He writes of poetical enthusiasm: Where the vivacity arises from a customary conjunction with a present impression; tho’ the imagination may not, in appearance, be so much mov’d; yet there is always something more forcible and real in its actions, than in the fervors of poetry and eloquence. The force of our 30. See Bricke 1980, 121–22 (also cf. 30–31, 46–58); MacNabb 1951/1966, esp. 71–76; Pears 1990, 50–51; and Price 1969, 186–88. 31. This is not to imply that beliefs are dispositions only to vivacity; they are dispositions to a range of manifestations that include vivacity, assent, and a variety of verbal and nonverbal behaviors. 32. A disposition’s tendency to change abruptly is a matter of degree; a dispositional belief is therefore a sufficiently infixed disposition to vivacity.

hume on stability, justification, and unphilosophical probability

205

mental actions in this case, no more than in any other, is not to be measur’d by the apparent agitation of the mind. (631/1.3.10.10) Since quasi-beliefs can be as vivacious as beliefs, such notions as “force” and “feebleness” must, in these contexts, apply to the relevant dispositions. The point is addressed more extensively in the context of resemblance. The effects of resemblance depend on a principle that is “fluctuating,” so that its “influence is very feeble,” and it is “impossible it can ever operate with any considerable degree of force or constancy” (109). By contrast, “The relation of cause and effect has all the opposite advantages. . . . [E]ach impression draws along with it a precise idea, which takes its place in the imagination, as something solid and real, certain and invariable” (110). The dispositions that result from the relation of cause and effect, unlike those that result from resemblance and poetical enthusiasm, do not tend to be fleeting. What accounts for this difference? In describing the second of the two systems of belief, Hume writes that it is “by their force and settled order, [that ideas] arising from custom and the relations of cause and effect, . . . distinguish themselves from . . . other ideas” (T 108). It is custom or habit, and hence repetition, that produces dispositions that do not tend to change abruptly. Hume puts this point by saying that repetition (116, 121), or custom or habit (86, 225), infixes belief. The dispositions produced by resemblance and poetical enthusiasm are not infixed, as they do not arise from repetition. Lack of fixity is a species of instability that is due entirely to the character of the mechanism that produces the state—the absence of repetition. Resemblance can intensify belief in that it can “assist that of cause and effect, and infix the related idea with more force in the imagination” (109), but cannot itself infix belief.33 According to Hume, dispositions that are not infixed do not qualify as beliefs (96, 453, 629/1.3.7.7).34 (In discussing fixity Hume often writes that a belief is an “idea” that is infixed. These are typical passages where Hume does not exercise care in distinguishing dispositions from their manifestations.) Hume’s tendency to suppose that we are introspectively aware of mental dispositions can obscure the role of dispositions and their properties in his analysis of belief. It is notorious that Hume claims that “we immediately feel a determination of the mind” (T 165) to associate a present impression with a lively idea of its usual attendant. This tendency is at work in the present context: ’[T]is impossible [the relation of resemblance] can ever operate with any considerable degree of force and constancy. The mind forsees and 33. Hume should say that when the relation of resemblance intensifies a belief, the result, strictly speaking, is a quasi-intensification of belief. 34. My discussion of the difference between belief and quasi-belief has been influenced by that of MacNabb 1951/1966, esp. 71–81, though I extend his treatment to poetical enthusiasm. MacNabb does not discuss the latter topic. The term ‘quasi-belief’, in this context, is MacNabb’s (79).

206

reflection and the stability of belief

anticipates the change; and even from the very first instant feels the looseness of its actions, and the weak hold it has of its objects. . . . [T]his imperfection is very sensible in every single instance. (109–10) A poetical description . . . may seem to set the object before us in more lively colours. But still the ideas it presents are different to the feeling from those, which arise from the memory and the judgment. There is something weak and imperfect amidst all that seeming vehemence of thought and sentiment, which attends the fictions of poetry. (631/1.3.10.10) Hume is describing what he takes to be a feeling or awareness of the disposition to vivacity produced by resemblance and poetical enthusiasm. This disposition is not infixed, and hence “loose” (106, 116, 123, 595, 624, 625), so that the mind does not have a “fast” (625, 627) hold on its object. This looseness is an “imperfection” in comparison to the fixity of dispositions that arise from the relation of cause and effect. Hume claims that we feel this imperfection in the disposition. It does not follow that Hume assimilates the disposition to an occurrent state. Hume’s distinction between beliefs and quasi-beliefs does not settle the question of whether there are justified quasi-beliefs. I do think Hume holds that quasi-beliefs are not justified, that in addition to not being infixed, they tend to be in disequilibrium. Here “general rules” come into play. Hume writes of poetical enthusiasm: “We shall afterwards have occasion to remark both the resemblances and differences betwixt a poetical enthusiasm, and a serious conviction. In the mean time I cannot forbear observing, that the great difference in their feeling proceeds in some measure from reflection and general rules” (T 631/1.3.10.11). Hume has told us in the preceding paragraph that we feel “something weak and imperfect amidst” poetical enthusiasm. The imperfection we feel is the looseness or lack of fixity, that is, the instability of the disposition. The difference in feeling between belief and poetical enthusiasm is “in some measure” due to a general rule. This raises the possibility that general rules contribute to the feeling of imperfection by increasing the instability of the disposition. Hume is more explicit that this possibility is realized in the case of the relation of resemblance: “And as this imperfection is very sensible in every single instance, it still increases by experience and observation, when we compare the several instances we may remember, and form a general rule against the reposing any assurance in those momentary glimpses of light, which arise in the imagination from a feign’d resemblance and contiguity” (110). In the context of the fourth kind of unphilosophical probability, general rules are (general) second-order beliefs, (general) beliefs about beliefs. In the present context, general rules are (general) beliefs about quasi-beliefs. For ease of exposition, I call both (general) beliefs about beliefs, and about quasi-beliefs, second-order beliefs. As an empirical fact, increases in vivacity that follow the feeling of imperfection are fleeting. The secondorder belief is a general rule. If the general rule heightens the feeling of imperfection, it must be because it increases the instability of the disposition.

hume on stability, justification, and unphilosophical probability

207

How is this accomplished? My treatment of this question elaborates an account due to MacNabb.35 Let us say that both beliefs and quasi-beliefs are putative beliefs. Consider the following set of doxastic states, where quasibeliefs, as well as beliefs, are counted among doxastic states: (a) one has a putative belief that p; (b) one believes that this putative belief follows the feeling of imperfection; (c) one believes that putative beliefs following feelings of imperfection are fleeting; and (d) one believes that one’s present putative belief that p will be fleeting. The belief at (d) follows from (b) and (c) by causal inference. Hume’s claim, I suggest, is that this set of doxastic states is, as an empirical fact, in disequilibrium. One believes that one’s present putative belief that p will be fleeting, and when one considers that one will soon abruptly let go of the putative belief, one is (even) less inclined to maintain the putative belief.36 The association of ideas by resemblance produces dispositions to vivacity that lack fixity, so that the putative belief that p, (a), is likely to change abruptly. The instability in (a) is due entirely to characteristics of the mechanism that produces (a), to the absence of repetition in the association of ideas by resemblance. In the absence of (c) and (d), however, there is no tendency for (a) to change owing to disequilibrium in the set of doxastic states. By contrast, there is disequilibrium, and hence an additional tendency to change, in the presence of the second-order beliefs (c) and (d). Consideration of the set of doxastic states that includes (c) and (d) inclines one to revise the set by relinquishing (a). In other words, the second-order beliefs render (a) even less stable, more imperfect. In the presence of (c) and (d), there is disequilibrium that magnifies the instability that is already in place due to (a)’s lack of fixity. When Hume states at page 110 that “we compare the several instances we may remember,” we need not take the comparing of instances to be a selfconscious activity. As feelings of looseness are observed to be followed by putative beliefs that are fleeting, we form the second-order belief (c). This belief results from custom. Similarly, when Hume states that we “form a general rule against the reposing any assurance in those momentary glimpses of light, which arise in the imagination from a feign’d resemblance,” we need not take the “general rule” to be a regulative principle governing belief formation. The presence of a feeling of looseness triggers the second-order habit, and we form the belief (d), the expectation that the assurance that p will be 35. See MacNabb 1951/1966, 96. 36. The case in which the relation of resemblance intensifies belief admits of an analogous treatment, by substituting “putative intensification of belief” for “putative belief” in (a–d), and in the accompanying discussion. In the case of poetical enthusiasm, the destabilizing secondorder belief has a different content. Poetical enthusiasm can enliven any idea to the point of putative belief (see T 631–32/1.3.10.10). Consider the following doxastic states: (a′) one has the putative belief that p; (b′) one believes this putative belief arises from poetical enthusiasm; (c′) one believes poetical enthusiasm can enliven any idea to the point of constituting a putative belief. Believing that one’s psychological state is equally conducive to putative belief in any proposition, one is less inclined to have the putative belief that p.

208

reflection and the stability of belief

fleeting. Consideration of the set of doxastic states that includes this belief generates disequilibrium. Since the belief at (c) is infixed, disequilibrium can best be removed by relinquishing the loose disposition that constitutes putative belief. Similar remarks apply to the case of poetical enthusiasm. This completes my interpretation of Hume’s resources for explaining why quasi-beliefs are not justified. The critical element in the account consists in locating second-order beliefs that generate disequilibrium. These second-order beliefs are instances of general rules.37 I am now in a position to offer an explanation of why Hume considers variations in degree of assurance in the context of the first three kinds of unphilosophical probability to be unreasonable. Hume must take it that these variations are sources of disequilibrium. On the basis of experience, we acquire the second-order belief that degree of assurance diminishes as psychological distance increases. Suppose one believes the following: (a) that some proposition p is true, where one holds that belief with degree of assurance k; (b) that one’s belief that p arises from memory or causal inference; (c) that degree of assurance diminishes as psychological distance increases; and (d) that one’s present degree of assurance that p would vary with changes in psychological distance. The belief at (d) follows from (b) and (c) by causal inference. The Humean claim is that this doxastic system is, as an empirical fact, in disequilibrium. When one considers that one’s degree of assurance in p would be diminished if the distance were less, and would be increased if the distance were greater, one is less inclined to maintain one’s present degree of assurance in p. (It is perhaps clearer that disequilibrium is present if the role of (c) is assumed by (c′): that large changes in degree of assurance accompany large changes in psychological distance.) In the case of quasi–beliefs produced by resemblance, the second-order beliefs generate disequilibrium that magnifies an existing instability. We have the same result in the context of the first two kinds of unphilosophical probability, where p is a memory belief. In the absence of beliefs (c) and (d), one’s doxastic system is unstable in that it will gradually change. As the event one remembers becomes more remote, one’s degree of assurance in the memory will diminish. This instability results entirely from the mechanism that sustains (a), from associationist degradation. In the absence of (c) and (d), however, there is no disequilibrium in the set of doxastic states. By contrast, there is disequilibrium, and hence an additional tendency for (a) to change, in the presence of the second-order beliefs (c) and (d).38 37. I do not claim to offer a fully general account of “general rules” in the Treatise. 38. Matters are somewhat different in the context of the third kind of unphilosophical probability. Once a particular causal inference, with the conclusion that p, is embedded in an argument someone produces, the degree of assurance in p is fixed. The degree of assurance is a function of the number of preceding steps. The addition of new steps does not change the length of the argument. The variations in degree of assurance in p are strictly hypothetical; one’s degree of assurance in p would have been diminished (increased), had the inference come after a longer (shorter) series of steps. Disequilibrium results from the consideration of these hypothetical differences, not (as in the case of memory) from consideration of variations in degree of assurance that in fact occur. There is disequilibrium, but it does not magnify an instability already in place.

hume on stability, justification, and unphilosophical probability

209

My explanation of Hume’s reasons for regarding the first three kinds of unphilosophical probability to be unreasonable assigns a critical role to second-order beliefs or general rules. There is a textual basis for this interpretation, even though Hume gives no examples of second-order beliefs, and makes no mention of general rules, in the course of discussing the first three kinds of unphilosophical probability at pages 143–46. Hume is explicit in assigning a role to general rules in his treatment of resemblance at page 110 of I.iii.9 and of poetical enthusiasm at page 631/1.3.10.10. In the final paragraph of this portion of the Appendix, Hume writes: “A like reflection on general rules keeps us from augmenting our belief upon every encrease of the force and vivacity of our ideas” (T 632/1.3.10.12). Hume directs us to insert this discussion of poetical enthusiasm at page 123 of I.iii.10. The discussion of the first three kinds of unphilosophical probability begins three sections, and twenty pages, later. (The intervening sections, I.iii.11–12, are devoted to philosophical probability.) Hume’s comment in the Appendix invites us to import general rules into contexts involving variations in vivacity due to sources other than resemblance and poetical enthusiasm. Furthermore, Hume explicitly invokes general rules in his discussion of the fourth kind of unphilosophical probability, and concludes his discussion of the two influences of general rules: “The following of general rules is a very unphilosophical species of probability; and yet ’tis only by following them that we can correct this, and all other unphilosophical probabilities” (150, emphasis added). General rules are implicated in the first three kinds of unphilosophical probabilities, as well as in the fourth. Hume does not, at pages 143–46, attempt to explain how the disequilibria associated with the first three kinds of unphilosophical probability are removed. I have recorded Hume’s contradictory and ambivalent assessments of a variety of epistemological distinctions in §§2–3 and offered an explanation of these tensions with reference to a two-stage model of Hume’s development of his theory of justification. At the close of §3, I suggested that, in his discussion of the fourth kind of unphilosophical probability, Hume is less concerned to document the claim that the instabilities in that context are eliminable, than to look ahead to the destructive stage by insinuating the demise of the project of justifying beliefs with reference to doxastic stability. Similar considerations apply to Hume’s discussion of the first three kinds of unphilosophical probability. The discussion concludes at page 146, four pages prior to Hume’s offering “a new and signal contradiction within reason” on behalf of “the skeptics.” There is a part of Hume that is pulled toward the epistemological distinctions to which he calls our attention, and hence to constructing a theory of justification that sustains those distinctions. There is also a part of Hume, as evidenced in his premature readiness at page 150 to suggest a “contradiction” internal to causal inference, that is pulled toward the destructive result that such distinctions cannot be sustained. In I.iii.13, this second part of Hume

210

reflection and the stability of belief

diverts him from attending to details of the constructive applications of his theory.39 How are the instabilities associated with the first three kinds of unphilosophical probability to be removed? When Hume uncovered a similar source of disequilibrium in connection with quasi-beliefs, he claimed that we restore equilibrium by relinquishing the doxastic states resulting from resemblance. In connection with unphilosophical probability, however, Hume cannot intend that we relinquish beliefs based on memory and causal inference. First, Hume holds that some beliefs based on causal inference (T 128, 147, 225) and memory (31, 153) are irresistible. Second, insofar as we are able to relinquish beliefs based on memory and causal inference, the result would be to sacrifice beliefs that comprise the two systems of realities, discussed at pages 107–8.40 These beliefs, attributed to the “judgment” and “understanding,” are preeminently reasonable at I.iii.9; thus, Hume does not, at I.iii.13, claim that any belief based on memory and causal inference is unphilosophical. It is the variations in degree of assurance, in the company of relevant second-order beliefs, that are the source of disequilibrium, and hence unreasonableness. This leaves Hume with the problem of explaining how equilibrium can be achieved in the context of the first three kinds of unphilosophical probability. I suggest that a solution must take note of the fact that the variations in degrees of assurance result from the mere passage of time or the mere length of argument. It is an empirical question whether long-term memory is less reliable than short-term memory, and whether lengthy arguments are less reliable than short ones. Suppose one has established empirically that one’s long-term memories are more often mistaken than short-term memories, or that one makes mistakes in conducting lengthy arguments more often than in conducting short arguments. Such correlations would support a lower degree of assurance in long- than in short-term memory, and in lengthy than in short arguments, respectively—one has established that the cognitive processes 39. There is another reason why Hume neglects his constructive project at I.iii.13. The third kind of unphilosophical probability suggests a special difficulty. A chain of causal inferences concerning ancient history would be “of almost an immeasurable length,” so that the vivacity of the original impression “must at last be utterly extinguish’d” (T 145). Yet historians do have beliefs about ancient history. Hume devotes as many pages (two) to this objection as to the description of the first three species of unphilosophical probability. Hume has two projects going at once: confirming his theory that belief is nothing but a lively idea, by explaining how belief arises associationistically; and providing an account of epistemological distinctions among beliefs. The first of these projects is primary and overwhelms the second project in the discussion of the third kind of unphilosophical probability. 40. MacNabb overlooks this difficulty (1951/1966, 96). He sees that Hume takes the position that “not expecting [beliefs based on resemblance] to remain beliefs, we cease to believe them.” More generally, we reject beliefs based on both resemblance and education in favor of “sticking to experience.” MacNabb states: “In the section ‘Of Unphilosophical Probability’ [Hume] deals with various other phenomena on the same lines.” These phenomena cannot be treated exactly “on the same lines”; unphilosophical probability does arise from experience— from perception, memory, and causal inference.

hume on stability, justification, and unphilosophical probability

211

involving greater psychological distance lead to an increase in the frequency of mistakes. Equilibrium can be achieved if one possesses second-order beliefs about the reliability of the relevant cognitive processes, and proportions one’s degree of assurance in a belief to the degree of reliability one attributes to the mechanism one takes to have produced the belief. Variations in degree of assurance will remain, but the variations will be supported by second-order beliefs about differences in degree of reliability. What is the mechanism that enables one to adjust degree of assurance in this way? Consider a case where the degree of assurance in the absence of a relevant second-order belief diverges from the degree of assurance that would be supported by the second-order belief.41 We can suppose that when the relevant second-order belief is present, it acts to weaken or strengthen the degree of assurance in the relevant first-order belief. I think this is a Humean solution. Hume’s argument in I.iv.1 that all knowledge degenerates into probability assigns a prominent role to empirical generalizations about the reliability of cognitive processes in regulating degrees of assurance. His reference to the “intricacy” (T 180) of a question and his explicit discussion of lengthy computations (181) show that he has in mind, in part, differences in reliability associated with the length of a demonstration. Hume’s argument (at pages 181–83) to show that all judgments of probability themselves degenerate into the absence of any belief also assigns a crucial role to generalizations about the reliability of cognitive processes in regulating assurance. In addition, we have seen (§3) that second-order beliefs or habits about the truth of classes of first-order beliefs weaken or strengthen first-order habits.

5. Concluding Remarks In this study, I have provided some evidence that Hume explicates justified belief with reference to doxastic equilibrium, and that he develops this theory in two stages. I have discussed a number of constructive applications of the theory: to beliefs based on the observation of accidental constant conjunctions; to doxastic states that arise from resemblance and contiguity, and from poetical enthusiasm; and to variations, due to psychological distance, in degree of assurance based on memory and causal inference. I believe Hume explores other constructive applications of his theory: to beliefs based on “education” (T 115–17); to beliefs based on the propensity to attribute

41. There is the possibility that the divergence between the degree of assurance and reliability at different distances is modest. For example, we have less assurance in long-term than in short-term memories. This is as it should be, if long-term memory is less reliable than shortterm memory. Evolution has perhaps provided a rough match between psychological degradation in assurance and cognitive degradation in reliability. Variations in degrees of assurance that generate disequilibrium in the absence of second-order beliefs about the reliability of cognitive processes might be stable in the presence of such beliefs.

212

reflection and the stability of belief

identity to related objects (198–210, 219–25, 253–58); to the distinction between principles that are “permanent, irresistable, and universal” and those that are “changeable, weak, and irregular”; and to moral or evaluative judgments based on the operation of sympathy (580–85).42 I have also called attention to the destructive application of the theory at I.iv.7. In my view, Hume’s pessimism in this section is hasty and overreaching.43 It is in part for this reason that I take the constructive applications of the theory to warrant detailed attention. My orientation to Hume’s epistemology is markedly different from the Kemp Smith interpretation, counterexamples to that interpretation aside. On the Kemp Smith interpretation, Hume’s theory of justification is entirely negative in its import; Hume simply makes the negative observation that it is pointless or false to claim that we ought not hold various beliefs. The interpretation provides no positive sense in which we ought to hold any belief. On my interpretation, Hume holds that there could be a positive reason for holding some beliefs, relative to stability as a cognitive objective. I defer questions about the foundation of Hume’s theory of justification to other occasions.44 I have sometimes said that Hume regards a set of doxastic states to be in disequilibrium, “as an empirical fact.” What is the Humean explanation of this empirical fact? The answer, I think, is closely related to another question that I have not pursued here. Why, in Hume’s view, is stability an objective of inquiry?45

42. I consider the second and third of these four examples in my 1991. (See note 7.) 43. Establishing this claim would require an extended discussion of I.iv.7, together with the material at I.iv.1 and I.iv.4 on which this section draws. 44. I begin to address such questions in my 1995b. 45. I am, as so often, indebted to David Velleman for detailed and incisive comments and discussion on earlier drafts. This study descends, in part, from my “Permanence in Belief in Hume and Descartes,” Hume Society, American Philosophical Association, Pacific Division Meetings, Portland, Oregon, March 23–26, 1988, and has benefited enormously both from the remarks of the commentator, Edwin McCann, and from the subsequent comments of David Hills. I have also benefited from discussion with John Immerwahr, Lawrence Sklar, and William Taschek. I read a version of this chapter at Union College and two related papers at the National Endowment for the Humanities Institute on David Hume and the Enlightenment, July– August 1990, Hanover, New Hampshire. I thank the participants—especially Charlotte Brown, Stanley Kaminsky, and Jan Ludwig—for helpful remarks on these occasions. Finally, I am grateful as well to referees for Journal of the History of Philosophy, and to the editor, for their comments.

hume on stability, justification, and unphilosophical probability

213

8 Hume’s Agent-centered Sentimentalism

might there be a principled compromise between the exacting selflessness of utilitarianism and the limited purview of egoism? Hume offers such a middle ground. In his theory of moral judgment, the variability in the intensity of sentiments produced by sympathy motivates the adoption of a steady and general point of view in evaluating an agent (§1). Hume’s appeal to a moral judge’s sympathizing with the members of the agent’s “narrow circle” does not fully rectify the variability in sentiment (§2). A complete development of Hume’s position requires identifying the moral point of view with the sympathetic perspective of the very agent whose action or character is under evaluation (§3). This perspective assigns value to the interests of persons in proportion to their closeness to the agent. The interests of the agent himself carry the greatest value, followed by those of his family, friends, and acquaintances; much less moral weight attaches to the interests of distant persons. A normative theory that is thus centered on the point of view of the agent is a natural consequence of Hume’s sentimentalism. This agent-centered theory explains aspects of the content, as Hume sees it, of our moral judgments: it requires considerable partiality to those with whom we have special relationships and only modest general benevolence (§4). This is a moral view worthy of consideration in its own right.

1. The Problem of Variation in Sentiment due to Sympathy Moral judgments, according to Hume, are “founded” (T 546; cf. 574, 608) in pleasurable or painful sentiments or feelings arising from sympathy with

214

“mankind” or “society” (578) insofar as its members are affected by the quality under evaluation.1 Sympathy is an associationist mechanism that converts the belief that a person is experiencing a particular feeling into the feeling itself (316–20, 385–86). We can relax this account; sympathy need only convert a belief in the existence of a feeling into a feeling with the same affective quality.2 Special resemblances and relations intensify the feelings sympathy produces (318). As a result, changes in a person’s relations to others lead to changes in the sentiments arising from sympathy, and differences in persons’ relations to others result in different sentiments. In light of the intrapersonal and interpersonal variations in sympathetic response, Hume raises an objection to his account of moral judgment: “But as this sympathy is very variable, it may be thought, that our sentiments of morals must admit of all the same variations. . . . But notwithstanding this variation of our sympathy, we give the same approbation to the same moral qualities in China as in England” (580–81); “The sympathy varies without a variation in our esteem” (581). Hume is sanguine, even cavalier: “to consider the case a-right, [the objection based on variation in sentiment] has no force at all; and ’tis the easiest matter in the world to account for it” (T 581). How can moral judgment, which is invariable, be founded in sympathy, which is variable in its effects? Hume’s answer is that “in order . . . to prevent those continual contradictions, . . . we fix on some steady and general points of view” (581–82). In a nutshell, the contradictions in judgment due to variation in sympathy motivate the adoption of a stable, and hence standard, point of view, that of a “judicious spectator” (581).3 There are two obstacles to Hume’s securing the steady and general point of view. In the first place, granting the need for a single perspective to serve as the standard in moral judgment, there is the problem of specifying the point of view that plays this role. Of the myriad sets of psychological relationships one person can bear to others, which set constitutes the standard perspective? The difficulty is to justify adopting a particular point of view as the standard.4 In the second place, there is the question whether Hume’s 1. [In this chapter, quotations of Hume are from the Selby-Bigge and Nidditch editions of the Treatise and second Enquiry.] 2. This suggestion is due to Stroud 1977, 197–98. 3. There is the question whether Hume can legitimately describe the variation in sentiment as involving “contradictions” (T 581; cf. 583, 602). The would-be contradictions vanish if superficially incompatible judgments are construed merely as statements of how virtuous or vicious a character appears from different perspectives. In providing an account of the motivation for adopting a standard point of view, Hume cannot presuppose the objectivity in moral judgment he seeks to explain. See Penelhum 1975, 143, and Mackie 1980, 121–23. One line of solution is that the variation in sentiment feels like contradictions and that the resulting conflict motivates its removal. For two versions of this solution, see Korsgaard 1999, esp. 14–17, 24–25, 35, 39n.24, and my 2002, 118–31. 4. There is an analogous distinction in the context of Hume’s political philosophy: “we must ever distinguish between the necessity of a separation and constancy in men’s possession, and the rules, which assign particular objects to particular persons” (EPM 310n.).

hume’s agent-centered sentimentalism

215

emphasis on a moral judge adopting a standard point of view, one that typically diverges from the judge’s actual point of view, is compatible with his root idea that moral judgment is founded in sentiment. If we adopt a standard point of view in order to remedy variation in sentiment, what remains of sentimentalism?5 In this essay, I focus on the first of these difficulties, though I consider its implications for the issue about the tenability of sentimentalism. I argue for a surprising, even paradoxical, Humean solution to the problem of fixing a standard point of view for moral judgment. This solution enhances the theoretical power of Hume’s sentimentalism: it enables him to explain why moral judgments are “disinterested” in a particular sense of the term; and it enables him to explain why the moral point of view requires only limited generosity to strangers, on the one hand, and also particular consideration for those to whom one bears special relationships, on the other. Within Hume’s sentimentalist framework, both disinterestedness and substantive moral principles can be derived from an independent solution to the problem of variation in sentiment. Let us consider the variation in sentiment due to sympathy in more detail. The strength of sympathy’s operation depends on our relationships to the persons with whom we sympathize. A variety of relationships matter: “We sympathize more with persons contiguous to us, than with persons remote from us: With our acquaintance, than with strangers: With our countrymen, than with foreigners” (T 581; cf. 318, 352, 582–83); “any peculiar similarity in our manners, or character, . . . or language” (318) is also a factor, as are causation and hence blood relation (318, 322; cf. 11), so that we sympathize more with family or relations (cf. 352, 487–88). Spatial and temporal contiguity or distance are special cases of psychological distance, which varies in all these relationships (cf. 318).6 Thus Hume writes that “we correct the different sentiments of virtue, which proceed from its different distances from ourselves” (585). The different “distances” encompass all the kinds of relationships that affect the strength with which sympathy operates. Hume recognizes that “those of the same trade, profession, and even name with ourselves” thereby stand in a special relation to us (352); presumably, we sympathize more with persons to whom we are related in these ways. Hume’s appeal to sympathy—the conversion of a belief in the existence of a feeling into a feeling with the same affective quality—is but one possible

5. For the difficulty, see Stroud 1977, 189–92; Snare 1991, 302; and Cohon 1997, 827–50. Cohon also offers a solution. For discussion and additional references, see my 2002, 133–36. 6. Spatial and temporal proximity and other special relations in Hume’s discussion correspond to what Hoffman calls a “here-and-now bias” and a “familiarity bias,” respectively, in empathic response. See his 2000, 13–14. Hume sees these features of sympathy as special cases of a general phenomenon.

216

reflection and the stability of belief

explanation of empathy.7 Here I follow recent psychological literature in using the term ‘empathy’ broadly, to include any vicarious affective response, affective responses more congruent with another person’s prior state or situation than with that of the person exhibiting the response.8 Sympathy, as Hume conceives it, is but one of numerous mechanisms—including, for example, mimicry and imaginative projection—that have been proposed to explain empathy.9 It is uncontroversial, however, that there exist the general kinds of empathic bias to which Hume calls attention.10 Three stretches of the Treatise address the problem of variation in sentiment due to sympathy. In the opening pages of III.iii.1, “Of the origin of the natural virtues and vices,” Hume argues that moral distinctions arise from sympathy (T 574–80). He pauses to “observe two remarkable circumstances in this affair, which may seem objections to the present system” (580). The first of these objections is that variability in sympathy is not matched in variability in moral judgment (580–81). The second is that we allow that someone can be virtuous even if circumstances impede the person’s character from having beneficial effects on others. Since “Virtue in rags is still virtue” (584), moral assessments cannot depend on sympathy with a character trait’s actual effects on society. The objection that sympathy is variable receives an extended discussion at pages 580–84. This material itself divides into two stages. At pages 581–83, Hume considers the possibility that adopting a standard point of view consists in correcting the sentiments themselves. At pages 583–84, he takes a second run at the problem, in light of the fact that sentiments often resist correction: we must be content with correcting our moral judgments rather than the sentiments in which they are founded. Hume discusses the second objection—virtue in rags—at pages 584–86, a discussion that draws on the theme of corrections. At pages 590–91, the penultimate paragraph of III.iii.1, Hume again addresses the variability in sympathy in the course of “a general review of the present hypothesis” (T 590), that moral distinctions arise from sympathy. There is a final treatment at pages 602–3, the second paragraph of III.iii.3, “Of goodness and benevolence.” In sum, Hume treats variation due to sympathy in an extended, two-stage discussion at pages 580–84 in III.iii.1, in a review of that discussion at pages 590–91, and at pages 602–3 of III.iii.3.

7. In Hume, the term ‘sympathy’ does double-duty: sometimes, it refers to the associationist mechanism introduced in II.i.10; other times, it refers to behavioral and psychological phenomena it serves to explain. Hume does not take care to flag his terminology with respect to this distinction. Doing so is perhaps unnecessary, since the first, technical sense is introduced in II.i.10 and maintains its explanatory role throughout Book III. 8. Cf. Hoffman 2000, 29–30, 93, and Preston and de Waal 2002, §1.1.3. 9. Hoffman reviews a variety of explanations for empathy (2000, ch. 2). For an updating of Hume in light of simulation theory, see Gordon 1995. 10. For a review of the empirical evidence, see Hoffman 2000, 61–62, 206–13, 251–52. For additional references, see Preston and de Waal 2002, table 1.

hume’s agent-centered sentimentalism

217

Before trying to flesh out the steady and general point of view, we need to consider Hume’s conception of how one arrives at corrected moral judgments. To adopt a standard point of view is to bring one’s evaluations of character in alignment with the sentiments one would have were one to sympathize from a standard psychological distance. Hume introduces an analogy to aesthetic judgment: ’[T]is evident, a beautiful countenance cannot give so much pleasure, when seen at the distance of twenty paces, as when it is brought nearer us. We say not, however, that it appears to us less beautiful: Because we know what effect it will have in such a position, and by that reflexion we correct its momentary appearance. (T 582) In the next paragraph, Hume observes, “Such corrections are common with regard to all the senses” (582). Elsewhere, he elaborates an analogy to perceptual judgments about size: The case is here the same as in our judgments concerning external bodies. All objects seem to diminish by their distance: But tho’ the appearance of objects to our senses be the original standard, by which we judge of them, yet we do not say, that they actually diminish by the distance; but correcting the appearance by reflexion, arrive at a more constant and establish’d judgment concerning them. (603; cf. 632/1.3.10.12 and EPM 227–28) In the moral case, changes and differences in relationships to others are sources of variability in sympathetic reactions to them. As I have noted, Hume holds that we are often unable to adjust our sentiments to conform to the ones we would have were we to sympathize from the standard point of view. (Hume makes this point not only at Treatise 582 and 583 in the course of the main discussion of variability but also at page 585 in the remarks about corrections that address the problem of virtue in rags, and again at page 603 in III.iii.3.) We correct our judgments, so that they reflect the sentiments we would experience were we to sympathize, from the standard point of view, with persons affected by the quality under evaluation.11 The theoretical issue is to identify the standard perspective or point of view from which we sympathize, to identify sympathy’s perspective, in the steady and general point of view.12 This is a slightly misleading shorthand,

11. I say “reflect” in order to maintain neutrality on the question of whether we best construe Hume as taking moral judgments to report (T 469) or express (471) underlying sentiments. 12. A number of passages suggest that corrections are achieved by comparing the sentiments the qualities of two persons would produce were we to adopt the same psychological distance in sympathizing with their effects (T 582, 584; EPM 227–28). This comparative

218

reflection and the stability of belief

in that we need not actually sympathize from this perspective. Putting matters more strictly, we tailor or align our moral judgments so that they reflect hypothetical sentiments, the sentiments we would have were we to sympathize from the standard point of view. In referring to this standard perspective, I write interchangeably of the psychological distance from which we sympathize in the steady and general (or moral) point of view, the psychological distance from which a moral judge sympathizes, and so forth. There is also the question of how we gain access to the hypothetical sentiments. Adopting the standard point of view is not a matter of sympathizing hypothetically, as if we were in the standard point of view when in fact we are not. Humean sympathy does not allow the sort of imaginative leap involved in taking up another’s point of view. Such simulation is more characteristic of Adam Smith’s account of sympathy than Hume’s.13 As I understand Hume’s view, our access to the sentiments depends on ordinary inductive inference. We know our own feelings that result from sympathy with those affected by the agent, sentiments that arise from our personal or individual perspective. Suppose we also know what that perspective is; in other words, suppose we know the nature of our special relationships to the persons affected by the agent. Finally, suppose we know, at least in a roughand-ready way, the psychological principles governing the correlation between sentiments arising from sympathy and psychological distance. If these conditions are satisfied, we are in a position to infer what our sentiments would be, were we to occupy the standard point of view. (But we do not thereby experience those sentiments.) This is no different from inferring how large an object would appear, from a standard perspective, on the basis of our knowledge of how it does appear from our individual viewpoint, how we are situated relative to the object, and the principles or laws of visual perspective.

account finesses the problem of selecting a standard psychological distance. Although Hume does write, “We naturally judge of every thing by comparison” (T 557; cf. 593–94), I do not see how he could rest with the comparative account. First, Hume thinks that “degrees” (547) of virtue and vice can be assessed noncomparatively: “there is just so much vice or virtue in any character, as every one places in it” (546–47), depending on the intensity of the relevant pain or pleasure (cf. 546). Similarly, Hume often writes without qualification of pronouncing a character virtuous or vicious, suggesting that a character is virtuous or vicious tout court (cf. 574–75, 591, 614). In the comparative account, corrected moral judgments could not take this form. Second, within the comparative account, the problem of variability due to differences in psychological distance reemerges as a problem of variability due to differences in points of comparison. To achieve comparability in our judgments, we would need to fix on the qualities of a single person or set of persons to anchor the comparisons. In evaluating Hume’s servant, on whom might we all fix? It is difficult to see how comparisons could converge even on the bestknown historical figures. The comparative account surfaces owing to an expository advantage: it enables Hume to introduce the idea of correcting judgments to reflect hypothetical sentiments, but without committing him to a specification of a standard psychological distance. 13. Cf. Darwall 1998, 264–65, 267, and 1999, 143–44.

hume’s agent-centered sentimentalism

219

Before considering how to specify the psychological distance from which we sympathize in the moral point of view, I take note of an apparent obstacle for any interpretation that identifies moral sentiments with hypothetical sentiments. According to Hume, moral duty or obligation requires “some actuating passion or motive, capable of producing the action” (T 518; cf. 517, 519, 532). Pleasures and pains arising from sympathy provide these motives, in light of a general desire for pleasure and aversion to pain (438). If to adopt the standard point of view is to align our judgments with the sentiments we would have were we to sympathize from a standard psychological distance, how could moral obligation secure its motivational foothold in actual sentiments of pleasure and pain?14 This is an acute form of the problem that Hume’s reliance on the steady and general point of view threatens to sacrifice to his underlying sentimentalism. My position is that this problem about moral motivation is genuine, but not as general as it seems. I defer this issue until §3, where I show that the most promising specification of the standard distance from which a moral judge sympathizes avoids the difficulty altogether. The solution to the problem of variability might seem straightforward: in order to remove variation in sentiment, moral judgments reflect the sentiments we would have were we to sympathize equally with everyone. This yields a utilitarian theory, giving equal moral weight to the pleasures and pains of every individual.15 Were we to sympathize equally with everyone, pleasures and pains would register equally with us, irrespective of whose pleasures and pains they are, and thus irrespective of differences in our relations to other persons. The present proposal must be understood in a particular way. There is no fixed set of circumstances we might occupy from which we could sympathize equally with everyone, anymore than there is some one location in the universe from which we could equally well perceive the size of every object. As we have seen, we sympathize with others differentially, depending on our relationships to them. There are no laws of psychology that enable us to infer what our sentiments would be, were we to sympathize equally with everyone from some particular historical position. Such a point of view is incompatible with the principles of psychology. To adopt a point of view in which we sympathize equally with everyone must be a matter of considering the sentiments we would have were we to sympathize with every other person from a perspective of the same kind, from the same psychological distance. The effect of this proposal is to ignore or bracket differences in relationships, without asking us to infer the result of a psychologically impossible feat of sympathizing equally with persons 14. On this point, see, in addition to the references at note 5, Harrison 1976, 111; Mackie 1980, 68; and Radcliffe 1991, 39. 15. Rawls associates classical utilitarianism with an impartial spectator “equally sympathetic to the desires and satisfactions of everyone affected by the social system” and attributes this position to Hume. See 1999, 163, 165, and 2000, 87. Harman also stresses the importance of impartiality in Hume and takes his theory to imply utilitarianism (1986, 1–3, 6, 9).

220

reflection and the stability of belief

with whom we have a variety of relationships. Rather, we align our moral judgments so that they reflect the sentiments we would have were we to sympathize with every member of society or mankind, from the same psychological distance in each case. Unfortunately, this version of the proposal does not solve the problem of variability. This is because there are a variety of ways to fix the standard kind of perspective from which we sympathize with each individual. One could take as the standard one’s relationship to an “indifferent” (T 488) person, or to an acquaintance, or to a friend, and so forth. The intensity of the resulting sentiments, and thus the degree to which a person is judged virtuous or vicious, will vary depending on which relationship serves as the standard psychological distance from which we sympathize. It is difficult to see why different moral judges should converge on a particular standard. Absent such convergence, variation in sentiment would remain.16 Granted, there are passages that one might try to give a utilitarian reading. Hume writes that a distinctively moral sentiment is one that results from “the mere view and contemplation” (T 475) or “the general survey or view” (614; cf. 574–75). These formulations, however, fall outside the passages in which variability is addressed. In any case, the entire problem is to understand what “the general survey” or “the mere view” consists in. Hume does provide one constraint on an account: moral sentiments arise “only when a character is considered in general, without reference to our particular interest” (472). This formulation requires abstracting from the special interests and relationships of the moral judge. It does not follow that moral judgment requires abstracting from particular interests and relationships altogether. (I return to this matter in §3.) To begin to see what Hume has in mind, we need to consider the passages that address variability in sympathy.

2. The Narrow Circle In order to solve the problem of variability in sentiment, Hume must specify, in a motivated way, the particular point of view that serves as the standard for moral judgment. The discussions of variability have a surprising feature: with but one exception (which I discuss in §3), they do not explicitly 16. I have deferred issues about reconciling Hume’s sentimentalism with his reliance on the steady and general point of view. It is nevertheless worth noting that this difficulty is severe if we are to sympathize with every person from a constant distance, as if we stood in the same relationship (e.g., friend) to everyone. On any account, a moral judgment will reflect a sentiment that somehow consolidates or combines the myriad of sentiments that arise from sympathizing with the different members of society. Contrived cases aside, a moral judge will stand in a particular kind of relationship to at most some of the persons affected by an agent’s character. So every moral judgment will reflect a set of sentiments, many of which are merely hypothetical. It is difficult to see how such a set could be consolidated into an actual sentiment in which a moral judgment is founded. The interpretation advanced in §3 has the virtue of grounding at least some moral judgments in actual sentiments.

hume’s agent-centered sentimentalism

221

identify sympathy’s perspective, the standard psychological distance from which a moral judge sympathizes; rather, they tend to focus on sympathy’s target or object, the persons with whom we hypothetically sympathize in the point of view that generates corrected or invariable moral judgments. On this point, Hume’s position is clear, both in the second stage of the extended discussion of variability and in the third and final discussion of the problem (§1): [W]e cannot . . . fix ourselves so commodiously by any means as by a sympathy with those, who have any commerce with the person we consider. (T 583, emphasis added) [W]e confine our view to the narrow circle, in which any person moves, in order to form a judgment of his moral character. When the natural tendency of [a person’s] passions leads him to be serviceable and useful within his sphere, we approve of his character, and love his person, by a sympathy with the sentiments of those, who have a more particular connexion with him. (602, emphasis added) In these passages, we encounter a doctrine of the narrow circle, the sphere in which a person moves or interacts (cf. 582, 590, 591, 602–3). As Hume states it, we peg our moral judgments to reflect the sentiments we would have were we to sympathize with the members of the agent’s narrow circle. It is possible, however, to sympathize with one and the same target from many different perspectives. The problem of variation in sentiment owing to variability in psychological distance calls for fixing a perspective, a point of view from which we sympathize in the moral point of view; instead, Hume tells us with whom we sympathize.17 What are we to make of this? Charlotte Brown and Christine Korsgaard provide the most sophisticated elaboration of Hume’s doctrine of the narrow circle.18 They take the general point of view to regulate sentiments about a person in two ways. First, we view the person, quoting Korsgaard, “through the eyes of our sympathy with herself and her friends, family, neighbors, and colleagues,” that is, by sympathizing with the person’s narrow circle.19 Second, we judge the person’s “characteristics according to the usual effects of characteristics of that kind, rather than according to their actual effects in this or that case,” that is, we 17. I know of only one discussion of this problem in the literature: Abramson 1999. Her solution is at 345–50. 18. At note 26, I recognize a reading of Brown’s position that diverges from Korsgaard’s in an essential respect. 19. Though we stand in special relations to persons in the same trade or profession (T 352), cases such as colleagues receive no play in Book III of the Treatise. I suspect Hume prefers to focus on cases where the groups with whom we sympathize differentially are nested, arranged in concentric circles in terms of psychological distance. (It is worth considering whether this simplifying assumption raises difficulties.) For the distinction between nested and overlapping circles, see Oldenquist 1982. For discussion, see my 2004, §I.3.

222

reflection and the stability of belief

appeal to “general rules.” This is uncontroversial as an account of what Hume says. The first regulatory device, sympathizing with the narrow circle, is prominent in Hume’s texts. The second regulatory device, general rules, is Hume’s answer to the objection that virtue in rags is still virtue (§1): “we . . . are delighted with the view of [a person’s] character, even tho’ particular accidents prevent its operation, and incapacitate him from being serviceable to his friends and country” (T 584). Hume’s response is, “Where a person is possess’d of a character, that in its natural tendency is beneficial to society, we esteem him virtuous” (584, emphasis added). A “general rule,” in this context, formulates the “tendency” (576, 577, 578, 579, 586, 602) of an agent’s underlying moral character or disposition to be beneficial to society. These “two regulating devices . . . constitute the general point of view.”20 The issue is whether the two regulatory devices are jointly sufficient to fix the general point of view. The difficulty for Hume’s doctrine of the narrow circle, as a solution to the problem of variation in sentiment, is that it tells us with whom we sympathize in the standard point of view but does not specify our relationships to those persons, and hence does not specify the perspective or distance from which we sympathize with them. The interpretation under consideration inherits this difficulty. The second regulatory device, general rules, is the only resource available in the interpretation to discipline the variation due to psychological distance. The texts suggest that Hume does not rely on general rules to solve the problem of variation in sentiment due to sympathy. Hume introduces this problem and virtue in rags as separate objections—“two remarkable circumstances” (§1).21 When Hume turns to virtue in rags at pages 584–86, his extended discussion of the first objection at pages 581–84 is behind him. Though he does return to variation in sentiment due to sympathy in the course of the “general review of the present hypothesis” at pages 591–92 and again in III.iii.3, he seems fully content with his earlier treatment. At pages 583–84, for example, he simply announces, “I now proceed to the second remarkable circumstance” (T 584), the point about virtue in rags and general rules.22 There is one way in which general rules might interact with the problem of variability in sentiment due to sympathy. General rules correct for idiosyncrasies

20. For the first two quotations from the literature, see Korsgaard 1999, 3, and 1996, 55; for the third, see 1999, 10; cf. 11–12. In each case, cf. Brown 1994, 23–25, and 2001, 197–98. 21. The two objections, however, can be seen as special cases of a general objection about variability: the first objection relates to variation in sentiment due to the fact that sympathetic response varies with psychological distance; the second also relates to variation in sentiment, but due to differences in the effects of a character trait in different sets of circumstances. 22. After appealing to general rules to solve the virtue in rags objection, Hume writes: “The case is the same, as when we correct the different sentiments of virtue, which proceed from its different distances from ourselves” (T 585). This passage perhaps encourages the interpretation under consideration. The cases are the same in that both require corrections to judgment, but there is no suggestion that the corrections for distance are themselves special cases of corrections for differences in the circumstances of persons under evaluation.

hume’s agent-centered sentimentalism

223

in the circumstances of the person under evaluation. If someone is virtuous, but down on his luck, accidental features of his situation interfere with virtue’s usual effects; general rules correct for such variation in “fortune” (T 585) by abstracting from these features. The question arises whether, in Hume’s view, the composition of a person’s narrow circle itself falls within the scope of the differences in circumstances addressed by general rules. A person who has familial virtues might have lost his family. Brown suggests that the moral point of view adjusts for a person’s “normal or usual narrow circle”; general rules correct for variation in the makeup of the narrow circle by fixing a canonical or standardized mix of family, friends, and acquaintances that is held constant in evaluating character, irrespective of the actual composition of the subject’s narrow circle.23 In this way, general rules impact the composition of the narrow circle itself. But even if general rules require sympathy with a standardized narrow circle, differences in the psychological distance from which we sympathize with its members will remain a source of variation in sentiment. This takes us back to the original problem. The textual curiosity remains in place: addressing the problem of variation in sentiment requires specifying sympathy’s perspective, a standard point of view from which we sympathize; Hume’s responses to the problem often seem misdirected, specifying instead with whom we sympathize, sympathy’s target. Even so, a doctrine of the narrow circle must constitute Hume’s solution to the problem. Hume raises the problem of variability at Treatise 580–81. The narrow circle passage at Treatise 583 is his response to it. All but one of the narrow circle passages fall squarely within one or another of the three discussions of the problem of variation; the sole exception—the passage at Treatise 590—immediately precedes one of these discussions. There is a related difficulty—beyond the misdirection—for Hume’s claim that a moral judge sympathizes with the agent’s narrow circle. What is the rationale for so restricting sympathy’s target? Korsgaard offers an answer, based on Hume’s observation in the second Enquiry that we approve of generosity or bravery even in our adversary (EPM 216). Korsgaard writes: I express antipathy for the enemy general who has captured me, perhaps castigating him as ferocious. The soldier guarding me, finding this grating, reminds me of the inspiration and other benefits he and his fellow soldiers gather from the very attribute I have just criticized. Sympathy with members of the general’s narrow circle presses me in the direction of admiring the attributes in question and so of admitting ‘courage’ as a virtue, regardless of its effect on my own interests. Since it is the members of his narrow circle who are most likely to have

23. Brown 1994, 24; cf. 25. The standard mix might vary, depending on the character trait in question. For related issues, see Abramson 1999, 343–44.

224

reflection and the stability of belief

sentiments about what a person does, it is from their point of view that such pressures and the attendant adjustments will most often be generated.24 .

Even granting the claims in the final sentence, there is no rationale here for sympathizing with the members of the narrow circle to the exclusion of those who fall outside it. They, too, may have sentiments, ones from a distinctive perspective, about the agent. Why do they not have a place in the mix? Indeed, Hume’s claim that a moral judge sympathizes with the agent’s narrow circle is in some tension with his view that sympathy operates in virtue of the resemblance among all humans: “There is a very remarkable resemblance [among all human creatures] . . . and this resemblance must very much contribute to make us enter into the sentiments of others” (T 318; cf. 359, 575). Any special relationship “facilitates the sympathy” (318). Granted, Hume might suppose that the intensity of sympathetic reactions diminishes exponentially with psychological distance—perhaps in accordance with an inverse square law—rather than in linear fashion.25 Whatever the function, “The sentiments of others have little influence, when far remov’d from us” (318). They have little influence, but not none at all. Psychological distance degrades sympathy’s effects, without eliminating them. When Hume says that in the moral point of view we sympathize with the agent’s narrow circle, this cannot mean that we sympathize with its members exclusively (602), on pain of sacrificing sympathy’s universal reach. Consideration of this tension enables us to understand the apparent misdirection in Hume’s response to the problem of variability. I suggest that Hume has an intended solution, which he does not state clearly. His intended solution is that to adopt the standard perspective is to sympathize from within the agent’s narrow circle.26 This allows, as Hume’s psychological doctrines require, that we sympathize with persons outside that circle, albeit less

24. Korsgaard 1999, 26; cf. 3, 15, 30–31. 25. Hutcheson had written in 1725: “This universal Benevolence toward all Men, we may compare to that Principle of Gravitation, which perhaps extends to all Bodys in the Universe; but, like the Love of Benevolence, increases as the Distance is diminish’d, and is strongest when Bodys come to touch each other. Now this increase of Attraction upon nearer Approach, is as necessary to the Frame of the Universe, as that there should be any Attraction at all. For a general Attraction, equal in all Distances, would by the Contrariety of such multitudes of equal Forces, put an end to all Regularity of Motion, and perhaps stop it altogether” (IMG §V.II). I owe this reference to Stephen Darwall. 26. Though Brown often writes of sympathizing “with” the members of the agent’s narrow circle (1994, 24, 25, and 2001, 198), it is possible that her intention is to attribute the intended solution to Hume. She does write that the narrow circle is the point of view “from which” we survey a person’s character (1994, 34n.12) and that we survey a person’s character “from the perspective” of that person’s narrow circle (24; cf. 2001, 198). Cohen and Garret interpret Hume along the lines of the intended solution; they respectively see Hume as focusing on the sentiment we would feel “were the action performed before our eyes and those involved close to us” (1990, 335) and “if one were well-placed in normal relations both to the person being evaluated and to those affected by her” (2001, 211). I do not find formulations of Hume’s intended solution in Korsgaard.

hume’s agent-centered sentimentalism

225

intensely than with those inside. Hume’s stated solution at Treatise 583 and 602 is that in the standard point of view we sympathize with the members of the agent’s narrow circle.27 This misleadingly implies that in the standard point of view we sympathize with the agent’s narrow circle exclusively. Although misleading, Hume’s stated answer to the problem of variability is an approximation that points in the right direction. In context, Hume must mean that a moral judge sympathizes most intensely with the members of the agent’s narrow circle. We could not achieve this intensity—given Hume’s claim that the effects of sympathy diminish with psychological distance—unless we sympathized with the members of the narrow circle from the vicinity of the narrow circle, either from within the circle or from nearby. (For simplicity, I follow Hume and focus on the former case.) Were moral judgments tailored to a point of view far removed from the narrow circle, the pleasures and pains within the narrow circle would not have any special influence. An answer to the question of the perspective from which a moral judge sympathizes thus falls out of the (misleading, though approximate) answer to the question of the persons with whom we sympathize, in conjunction with Hume’s account of the operation of sympathy. Though the gap between the stated and the intended solutions to the problem of variability obscures Hume’s underlying position, it seems harmless enough in and of it itself. We shall see in §3, however, that the gap interferes with Hume’s fully developing his solution to the problem of variability. Stepping back, two sources of variation in sentiment permeate Hume’s thinking. One source is differences in whose interests are taken into account: sympathy yields different results to the extent that we sympathize with the pleasures and pains of different persons. These are variations due solely to sympathy’s target or object. Hume has this source of variation in view when he emphasizes that in moral judgment we overlook our own interests (T 472). The second source is the variation in sentiment produced by sympathy due to changing and differing relations to the persons with whom we sympathize, variation in psychological distance. Even if moral judgment reflects sympathizing with a fixed group of persons, one’s interests—in the form of special relationships to family, friends, acquaintances, countrymen, et cetera—will yield variation in sentiment. Holding the target of sympathy fixed, variation in sentiment due to one’s interests enters through the back door, through the psychology of sympathy itself. In the exercise of sympathy, interests come into play twice over, once in connection with the object of sympathy, and again in connection with the operation of sympathy, given its object. 27. Frequently, commentators slide back and forth between interpreting Hume as advancing the intended and the stated solutions. Brown writes: “[W]e survey a person’s character from the perspective of that person’s narrow circle. . . . We sympathize with the . . . person’s narrow circle” (1994, 24, emphasis added; cf. 2001, 198). Swain writes: “The moral sentiments arise when we consider the situation from . . . the point of view of those who are affected by the person being judged” and “the general point of view . . . is achieved by sympathy with those affected by the person we are judging” (1992, 483, 484, emphasis added). Of course, the lack of clarity is due to Hume.

226

reflection and the stability of belief

But we need not say that Hume confused the questions of sympathy’s target and sympathy’s perspective. The steady and general point of view must remove both sources of variation. Hume’s psychological theory ensures that the answers to the two questions are interdependent, so that the two sources of variation are addressed in a unified way. Sympathy decays with psychological distance. Thus, to fix the perspective from which we sympathize in the moral point of view is to fix with whom we (most intensely) sympathize—as those nearest in psychological space to the specified point of view. And, as we have seen, to fix the group of persons with whom we (most intensely) sympathize is to fix the perspective from which we sympathize— as a point of view within that group. As a special case, to sympathize (most intensely) with the narrow circle is to sympathize from within the narrow circle. These interdependencies foster the impression that the answers to the question of target and to the question of perspective are the same. I have addressed the relationship between Hume’s stated and intended doctrines of the narrow circle. What is the rationale for the intended doctrine in the first place? In the extended discussion of the problem of variability, Hume calls attention to the intra- and interpersonal variation in sentiment “[w]hen we form our judgments of persons, merely from the tendency of their characters to our own benefit, or to that of our friends” (T 583). The variation leads us to “seek some other standard of merit and demerit, which may not admit of so great variation” (583). “Being thus loosen’d from our first station,” Hume writes, “we cannot afterwards fix ourselves so commodiously by any means as by a sympathy with those, who have any commerce with the person we consider” (583). Although Hume draws a conclusion about the persons with whom we sympathize, the questions of sympathy’s object and sympathy’s perspective are interdependent. Fixing ourselves by sympathizing from the perspective of the narrow circle is itself commodious, in the now archaic sense of convenient or serviceable. It is serviceable because of its salience; it is an obvious point of view on which we can all converge.28 A passage in the second Enquiry coheres nicely with the discussion at Treatise 583 of a commodious solution to the problem of variability. In context, Hume is addressing the question of whose interests are taken into account; the problem of variability in sentiment due to psychological distance is not officially in play. We know, however, that interests enter into the operation of sympathy itself, through the impact of special relationships. Hume writes: Usefulness is agreeable, and engages our approbation. This is a matter of fact, confirmed by daily observation. But, useful? For what? For somebody’s interest, surely. Whose interest then? Not our own only: For our

28. Sayre-McCord appeals to salience in interpreting Hume’s doctrine of the narrow circle, though his emphasis is on explaining why it is the members of the narrow circle with whom we sympathize in the standard point of view (1994, 217, 218–19).

hume’s agent-centered sentimentalism

227

approbation frequently extends farther. It must, therefore, be the interest of those, who are served by the character or action approved of. (EPM 218) What are the considerations that lead Hume to conclude in the final sentence that the interests that matter are those of the narrow circle? He does not fill in an argument. We can think of Korsgaard (who focuses on a passage two pages earlier) as trying to supply a rationale, discussed previously: the persons most likely to have feelings about an agent’s qualities are the members of his narrow circle. A more natural reading is that the interests of the narrow circle are especially salient. Hume often appeals to psychological salience in his political philosophy (cf. T 507, 510–13, 559). Rules governing stability in possession are “obvious” (493, 514). Rules of property, possession, and succession incorporate distinctions determined by “the imagination” (507, 513n.1) or by “the fancy” (508n.), and hence distinctions that are salient under the associationist principles that Hume takes to govern human psychology.29 In the context of Hume’s moral theory, where the problem is to fix a standard point of view for evaluating the qualities of an agent, the imagination will most readily run to the point of view of the agent’s narrow circle. We can suppose this has an associationist explanation, precisely because of the special relationships between the agent under evaluation and the members of his sphere.

3. The Agent-centered Solution to the Problem of Variability So far, so good. As it stands, however, the intended, implicit answer to the question of the perspective from which we sympathize in the steady and general point of view is inadequate. With the standard point of view identified as that of the narrow circle, a source of variation in sentiment remains. Members of the narrow circle—the agent’s family, friends, and acquaintances— bear different relationships to each other and to the agent. Just as it is psychologically impossible to sympathize equally with everyone universally, it is impossible for a member of the narrow circle to sympathize equally with all its other members. For example, not all the agent’s friends will be friends with each other, so that there will be differences in their sympathetic responses to one another. Sympathizing from within the narrow circle does not ensure sympathizing with the members of the narrow circle from a determinate psychological distance.30 Though Hume does not raise this difficulty, 29. For discussion of the role of the imagination in Hume’s political philosophy, see Harrison 1981, §3. 30. It might be suggested that Hume try to take the view that moral judgment is inherently indeterminate, so that resolution of the variation is not required. Perhaps, but the sources of variation in sentiment that arises within the narrow circle are of a piece with those that give rise to the problem of variability in the first place. Also, the variation could be extreme if the narrow circle includes rivals and adversaries. See §4.

228

reflection and the stability of belief

further resolution of this standard perspective is required to eliminate variability in sentiment.31 The members of the agent’s narrow circle stand in different relationships to each other. The resulting variation in sentiment can be removed only if we sympathize from the point of view of a homogeneous group of persons, a group whose members stand in the same psychological relationships to each other and to persons outside the group.32 Putting aside artificial cases we might construct, such homogeneity in perspective can be secured only by sympathizing from the point of view of a particular member of the narrow circle. On which point of view might we all fix? Consider Korsgaard’s example (§2), itself inspired by Hume: the soldier, the general’s chief of staff, and the general’s spouse have different relationships to the members of the general’s narrow circle, including different relationships to the general. The obvious solution is to identify the standard point of view with that of the general, the agent under evaluation. In order to remove the variation in sentiment within the narrow circle, the imagination, in evaluating the agent, will ultimately run to the point of view of the agent himself. I call this the agent-centered solution to the problem of variation in sentiment due to psychological distance; I call the resulting moral theory agent-centered sentimentalism.33 Hume comes close to stating the agent-centered solution to the problem of variation in the summary discussion of variability in III.iii.1. In considering this passage, it is important to recall that Hume misrepresents the doctrine of the narrow circle, as if it were a claim about the persons with whom we sympathize in the moral point of view rather than about the point of view from which we sympathize (§2). The summary discussion, however, is perhaps the only context in which Hume explicitly asks the question of the perspective from which we sympathize: “every particular person’s pleasure and interest being different, ’tis impossible men cou’d ever agree in their sentiments and judgments, unless they chose some common point of view, from which they might survey their object, and which might cause it to appear the same to all of them” (T 591, emphasis added). His answer, which follows immediately, begins: “Now, in judging of characters, the only interest or pleasure, which appears the same to every spectator, is that of the person himself” (591, emphasis added). If we construe this passage as answering the question Hume asks, the question of sympathy’s perspective,

31. In addition, different members of the agent’s narrow circle will move in different, albeit overlapping, spheres, so that they will sympathize most intensely with different groups of people, groups on whom the agent might have quite different effects. This problem is perhaps solved by the standardization, via general rules, of the composition of the agent’s narrow circle (§2). 32. Hoffman wrestles with the worry that an empathic morality “may be ideal in homogeneous groups” but not otherwise (2000, 216; cf. 246, 283, 295–96). 33. Brown uses the terminology of an “agent-centered theory” for a different purpose (1994, 20–21).

hume’s agent-centered sentimentalism

229

this is the agent-centered solution to the problem of variation.34 (I am not suggesting that the quotation requires this solution; the continuation of the passage, which I discuss later in this section, offers a different, more familiar, response.) Perhaps other devices could eliminate the indeterminacies inherent in identifying the standard point of view with that of the narrow circle, but it is difficult to envision a more elegant and salient solution.35 Solving the problem of variability requires that we align our moral judgments with the sentiments we would have were we to sympathize with society from the perspective of the agent himself. Since we evaluate many agents, strictly speaking we “fix on some steady and general points of view” (581–82, emphasis added and deleted) in our moral judgments. A further development of the agent-centered solution to the problem of variability is required if it is fully to eliminate variation in moral judgment due to psychological distance. I begin with a preliminary matter. Moral judgments extend to action, sentiment, and character (T 471, 475). Though the evaluation of character is fundamental (575), in the remainder of this chapter, I often conduct the discussion with reference to actions. This simplification smoothes the exposition. We can now call attention to a residue of variation in sentiment within the agent-centered solution. Even sympathizing from the agent’s point of view, sentiments will vary over time. The “incessant changes of our situation” (583) are one of the main sources of variation. The agent’s relationships to others, and to members of the narrow circle in particular, are susceptible to change: a friend becomes a spouse; an acquaintance becomes a friend. In addition, sentiments that assess the action an agent performs at a particular time will themselves vary with proximity to the time in question, and hence proximity to the action’s effects. (Hume is acutely aware of variation in sentiment due to differences in temporal contiguity and remoteness; in III.ii.7, such variations are the source of the necessity of civil magistrates.) The obvious solution is to fix the standard temporal distance for evaluating an action at zero; in evaluating an agent’s action at time t, we sympathize from the point of view of the agent himself specifically at t.36 This constitutes 34. But the passage is not unequivocal. It is possible to construe it as lapsing into an answer to the question of sympathy’s target—if we take it that the interest or pleasure of the agent “appears” to every spectator by way of sympathizing with it. 35. Much as sympathy produces variation in sentiment, memory and causal inference produce variation in degree of confidence. (See my 2002, 34, 105, 119.) For example, our confidence in memory beliefs diminishes as the matter of fact we remember becomes more remote or distant. Corrections are thus required in the context of “unphilosophical probability” at Treatise I.iii.13. I have argued that, in the case of memory, corrected judgments reflect the degree of confidence at the time the memory belief was first acquired, that is (we might say), “up close” to the event (2002, 137). The agent-centered solution is highly analogous in this respect. Comparison of Hume’s treatments of unphilosophical probability and moral sentiment is pursued, along different lines, by Wallace (2002). 36. There is perhaps some affinity here with the similarly paradoxical principles of dated rightness, principles that require indexing to the time of the action, developed by H. S. Goldman (1976).

230

reflection and the stability of belief

the steady and general point of view with which moral judgments are aligned. (What if the agent knows at t that an acquaintance at t will be a family member at a later time, t′? If so, at t, the agent will sympathize with the feelings of the person qua acquaintance and future family member.) In any interpretation, the variability in moral sentiments and associated “contradictions” in judgment motivate the adoption of a steady and general point of view. In a fully developed version of the agent-centered solution to the problem of variability, the standard for moral assessment of an action at t consists in the point of view of the agent at t.37 Moral judgment proceeds from the point of view occupied by the agent at the time of the action under evaluation. This account has an important implication for understanding how the agent, in the moral point of view, factors in his own pleasures and pains. Given Hume’s psychological theory, moral judgments reflect the sentiments that arise from sympathizing with everyone, though most intensely with the members of the agent’s narrow circle. The agent is himself a member of his narrow circle, not one to be ignored or discounted. How will the agent take into account the effects of his action on himself? Since he will experience such pleasures and pains firsthand, without needing to rely on sympathy, it might seem that sympathy is not required in this special case. This is not correct. Moral judgment of an agent’s action at t reflects the sentiment one would have were one to sympathize at t with the pleasures and pains that result from the action. These lie in the future, so that as of time t, the agent will not have (nonsympathetically) experienced any of those effects. He therefore must take into account the future effects of his action upon himself in the same way that he takes into account its future effects on others, by sympathizing with a future self (his own).38 In the steady and general point of view, the agent sympathizes with his future self, as well as with the future selves of others. It must be granted that Hume seems quite unable to hold on to the agentcentered solution to the problem of variability. The passage that suggests the solution—the one context in which Hume directly asks the question of sympathy’s perspective—quickly reverts to a doctrine of the narrow circle: “the only interest or pleasure, which appears the same to every spectator, is that of the person himself whose character is examin’d; or that of persons, who have a connexion with him” (T 591, emphasis added). Why does Hume

37. Appealing to Treatise 591, Stewart briefly considers, and rejects, the agent-centered solution: “Any particular man’s point of view, even the agent’s, is surely neither inalterable nor general” (1976, 179). The refinement I offer responds to the point that the agent’s point of view is not inalterable. Slightly later in her article, Stewart suggests that we understand the moral point of view as “general” in the sense that it abstracts from circumstances idiosyncratic to particular situations (182–83). The agent-centered solution can of course help itself to “general rules” in this sense. See later in this section. 38. Since sympathy varies with proximity in time, the agent will sympathize most intensely with his near-term self; there are implications here for a Humean account of prudence. For criticism of accounts “biased in favor of what is near,” see Kagan 1989, 280–83.

hume’s agent-centered sentimentalism

231

back off from the agent-centered solution in favor of the doctrine of the narrow circle? I believe we can locate an explanation that does not count against interpreting Hume in terms of the agent-centered solution. As we have seen (§2), in the moral point of view we sympathize from the perspective of a specified group of persons if and only if we sympathize (most intensely) with the persons in the specified group. Against this background, Hume has a propensity to formulate his position with reference to the question of sympathy’s target in the steady and general point of view. This is perhaps because he raises the issue of variation due to differences in whose interests are taken into account at Treatise 472, more than one hundred pages before introducing the problem of variation in sentiment due to psychological distance at 580–81. This is itself unsurprising. The question of whose interests we consider is a substantive matter that arises within any normative ethical theory; the question of the perspective from which we sympathize is distinctive to a sentimentalist framework that relies on empathic mechanisms to produce moral feelings. When Hume specifies the group with whom we sympathize in the steady and general point of view, he has a further propensity to omit the qualification “most intensely” in favor of an approximate, misleading answer to the question of sympathy’s target (§2). In light of the interdependence of the question of target and the question of perspective, an unqualified specification of a group of persons with whom we sympathize thus serves as the proxy for the specification of the point of view from which we sympathize. This goes smoothly enough within the context of the intended doctrine that we sympathize from the perspective of the narrow circle. The proxy for the doctrine is that we sympathize with the members of the agent’s narrow circle. By contrast, consider the agent-centered solution to the problem of variability, the position that we sympathize from the perspective of the agent who is under evaluation. The proxy for this position is that we sympathize with the agent who is under evaluation. But this statement—that in evaluating the agent we sympathize with the agent himself—seems hopelessly counterintuitive. It suggests that the moral point of view takes into account only the interests of the agent. This smacks of egoism, and thus obstructs Hume’s ability to maintain the agent-centered solution. The egoistic feel of Hume’s position, however, is an illusion. In the steady and general point of view, we sympathize with the agent. That is shorthand for the position that we sympathize most intensely with the agent. Given that sympathy decays with psychological distance, the point of view from which we sympathize most intensely with the agent must be that of the agent himself. Given sympathy’s universal reach, to sympathize from the point of view of the agent is to sympathize with society or with mankind, to sympathize with everyone, in varying degrees—most intensely with the agent, to a lesser degree with the other members of the agent’s narrow circle, and so forth. The role of sympathy ensures that the pleasures and pains that the agent’s action produces in everyone are taken into account. The agent-centered solution is thus not equivalent to egoism. Hume tends to characterize the

232

reflection and the stability of belief

steady and general point of view in terms of an unqualified answer to the question of with whom we sympathize. In the context of the agent-centered solution to the problem of variability, such a formulation appears egoist, without really being so. Hume’s focus on the unqualified answer to the question of with whom we sympathize stands in the way of a fully consistent development of his response to the problem of variability. Am I overlooking genuine obstacles to Hume’s adopting the agent-centered solution to the problem of variability? I will consider four objections.39 The first involves a residual source of paradox. It might seem that Hume cannot identify the standard psychological distance with that of the agent, simply because the agent will be the subject of a great many nonmoral sentiments. To identify the steady and general point of view with that of the agent, however, is not to incorporate any or all of the agent’s sentiments into the standard perspective. The agent-centered solution focuses on a particular sentiment of the agent, one that results from a specified process. Hume observes that “under the term pleasure, we comprehend sensations, which are very different from each other” (T 472). An agent will typically experience many painful and pleasurable sentiments, including such direct passions as joy, grief, sorrow, fear, hope, desire, and aversion (II.iii.9), as well as the indirect passions of pride, shame, love, and hatred (I.ii.1–2).40 These sentiments of the agent do not constitute the standard for moral judgment. A distinctively moral sentiment or feeling is, in Hume’s words, “a pleasure or uneasiness of a particular kind” (471, emphasis added; cf. 469, 475, 499). Hume is aware that it might be difficult to recognize this sentiment for what it is: ’Tis true, those sentiments, from interest and morals, are apt to be confounded, and naturally run into one another. It seldom happens, that we do not think an enemy vicious, and can distinguish betwixt his opposition to our interest and real villainy or baseness. But this hinders not, but that the sentiments are, in themselves, distinct; and a man of temper and judgment may preserve himself from these illusions. (472)

39. What if the agent under evaluation is excessively sensitive to the feelings of others? What if he is a psychopath with little or no capacity for empathy? Are such agents subject to correspondingly stringent and permissive moral requirements, respectively? Hume has some degree of freedom in dealing with this source of variation in sentiment. One option is to take the agent’s empathic capacity into account. Another is to appeal to general rules, relying on the empathic capacity of a normal person, or of a normal person in the circumstances. In any event, the variation in sentiment owing to differences in empathic capacity is a source of variability beyond that due to differences in psychological distance. Any solution to this latter problem will also have to address the question of how to calibrate the degree of sensitivity characteristic of the steady and general point of view. This is a general issue for sentimentalism, not for agentcentered sentimentalism in particular. 40. Even if, as a number of commentators hold, Hume takes moral approval and disapproval to be faint forms of love and hatred, many direct and indirect passions are not moral sentiments.

hume’s agent-centered sentimentalism

233

It is only the sentiment that arises within the agent from sympathizing with society or mankind, albeit more or less intensely depending on the distance of various others from the agent, that serves as the standard for moral judgment.41 The second objection is related to the first. Might a problem lie less with the agent’s recognizing distinctively moral sentiments than with his coming to possess them? In particular, it might seem that the agent-centered solution to the problem of variability makes it too easy for the agent to acquire a moral sentiment. After all, the standard point of view just is that of the agent. It remains the case that a moral sentiment results from the agent’s sympathizing, albeit from his own point of view, with all the pleasures and pains an action produces; the agent must achieve great sympathetic breadth. The operation of sympathy also requires the possession of appropriate beliefs or knowledge about the future condition of other persons.42 What is more, in morally evaluating his own actions, the agent is bound by the strictures of general rules (§2). The agent must sympathize with the usual or normal effects of his actions, with their effects in circumstances that typically diverge from the particulars of his situation.43 Acquiring a moral sentiment is a far from trivial exercise. The third objection relates to Hume’s views about moral psychology. For Hume, moral judgments are founded in sentiments that account for moral motivation; this claim seemingly presents a problem for any interpretation in which adopting the standard point of view consists in aligning our judgments with hypothetical sentiments, sentiments we would have were we to sympathize from a standard psychological distance (§1). Does not the agentcentered solution run afoul of the requirement that moral obligation derives from moral motivation? After all, on the agent-centered solution, we correct our moral judgments, though not our sentiments themselves, by aligning judgments with the sentiments we would have were we to sympathize with society from the agent’s point of view; we need not actually experience these sentiments. The agent-centered solution addresses this difficulty in a way that is uniquely fitted to Hume’s views about moral motivation.

41. Some commentators take moral sentiments to be phenomenologically distinctive. See, for example, Broad 1930, 84; Rawls 1999, 162, and 2000, 93; and Garrett 2001, 212–13. My claim that moral sentiments arise from a special kind of psychological process is compatible with this interpretation but does not require it. There is the possibility of misidentification even on the phenomenological interpretation. Cf. Stewart 1976, 185–86. 42. There is also the question of how much information about the typical effects of a character or action one must take into account in the moral point of view. For example, to what extent must a moral judge inform himself about human psychology? For some discussion, see Abramson 1999, 347–49, and my 2004, §§I.3–4. 43. The alternative would be to sympathize with the pains and pleasures that would be the actual effects of an action, given the particulars of the agent’s situation. The appeal to general rules, along the lines of Hume’s response to the virtue in rags objection, converts such actsentimentalism into rule-sentimentalism, a distinction paralleling that between act- and ruleutilitarianism.

234

reflection and the stability of belief

Though a moral judge need not experience a sentiment that would arise from the standard perspective, it is not a sentiment humans are precluded from experiencing. A moral judge could have the sentiment that results from sympathizing from the agent’s point of view, were the judge and the agent the same person.44 (At Treatise 603, Hume acknowledges that “tho’ this advantage or harm [of the narrow circle] be often very remote from ourselves, yet sometimes ’tis very near us, and interests us strongly by sympathy.” In other words, often we are close to the agent under evaluation; indeed, often we are the agent.) It is the agent who is most directly subject to moral obligations in light of moral judgments about him, and hence who needs to possess the necessary motivation or sentiment.45 If it is morally wrong for an agent to perform an action, then he ought not perform it. Of course, as we have seen, the agent will have many pleasurable and painful sentiments, beyond any distinctively moral sentiment. The motivational efficacy of these sentiments depends on their strength or weakness in the circumstances (cf. T 418–19, 437–38). The important point is that the agent-centered theory delivers moral sentiments that ground moral obligation in that they are available to the agent to compete with his other motives at the time the action ought to be performed.46 By contrast, suppose Hume were to identify the standard perspective with that of a third party—for example, a member of the agent’s narrow circle. On any such account, the sentiments that arise in the standard point of view would necessarily remain hypothetical for the agent and thus be deprived of their motivational function.47 These considerations about the relationship of the steady and general point of view to moral motivation suggest a fourth objection, related to Hume’s conception of disinterestedness (§1). When the agent and the moral judge are the same person, moral assessment reflects the sentiment arising in the agent from sympathizing with society from his own point of view. Since we sympathize more intensely with persons with whom we stand in close psychological relationships, it is the pleasures and pains of those with whom the agent has special relationships that receive special weight in moral

44. There could be divergence to the extent that the agent’s actual circumstances are different from those in the canonical or standardized set of circumstances that “general rules” bring into play. 45. Mackie observes that “The only sentiment that could directly influence action would be one which the agent himself actually had at the time of acting,” but does not take the opportunity to suggest the agent-centered solution (1980, 68). 46. The treatment of the issues here is distinct from that in my 2002 (135–36). Think of my earlier discussion as proposing the best solution I can muster to the problem about moral motivation and hypothetical sentiments in its general form. There is a much better solution in the special case where the standard point of view is identified with that of the agent. 47. The agent-centered solution thus solves the complaint, as Abramson puts it, that Hume’s judicious spectator theory “commits him to the view that a virtuous agent’s moral assessments ordinarily play no role in motivating his conduct” (2002, 301–2). The point of view of the judicious spectator coincides with a sympathetic perspective that is uniquely available to the agent. Abramson proposes a distinct solution.

hume’s agent-centered sentimentalism

235

judgment. Moral judgments are not disinterested in the sense of giving equal weight to everyone’s interests, as a utilitarian position would have it; moral judgments give special weight to the interests of the agent’s narrow circle. Does this result conflict with Hume’s insistence that “’Tis only when a character is considered in general, without reference to our particular interest, that it causes such a feeling or sentiment, as denominates it morally good or evil” (T 472)? The proper response is that moral judgment does give special weight to one’s own interests if one happens to be a member of the agent’s narrow circle; the moral point of view is disinterested—it proceeds without reference to one’s particular interests, if one is not a member of the agent’s narrow circle. Qua moral judge, we overlook our own interests; qua agent, our own interests figure prominently in moral judgment. This account of the role of interests in moral judgment is a consequence of the agent-centered solution, in which the agent’s point of view and the moral point of view coincide, to the problem of variability. It is interesting to consider Hume’s position against the background of Bernard Williams’ discussion of personal relations in a Kantian context. Williams considers a person who can save only one of two people in equal peril and chooses to save his wife: “It might have been hoped by some (for instance, by his wife) that his motivating thought, fully spelled out, would be the thought that it was his wife, not that it was his wife and that in situations of this kind it is permissible to save one’s wife”; as Williams puts it, the latter formulation “provides the agent with one thought too many.”48 Hume’s agent-centered theory eliminates the need for the extra thought. In forming a moral judgment, a person other than the agent will need to determine the sentiments he would have in the circumstances were he the agent and hence were he the woman’s spouse. The agent himself, however, need not go through any such step. He need only sympathize with those affected by his action, including his wife; when the agent does sympathize with his wife, the perspective of their special relationship is built into the operation of sympathy. Although the agent, in arriving at moral judgment about himself, will have to take care to sympathize with the whole of society and to make any allowances for general rules, he will not have to adopt a sympathetic perspective that diverges from the one he automatically occupies in virtue of the fact that he is himself the subject of evaluation. The agent’s perspective constitutes the moral point of view.

4. Normative Implications of Agent-centered Sentimentalism Having considered some main objections to interpreting Hume as an agentcentered sentimentalist, I turn to this position’s normative import. The 48. B. Williams 1981, 18.

236

reflection and the stability of belief

agent-centered characterization of the steady and general point of view generates consequences that have an affinity with the agent-centered restrictions that some have sought to introduce into utilitarian moral theories. Similarly, Hume’s theory enables him to explain both why morality imposes strengthened moral obligations to those with whom we have special relationships and why morality requires only limited generosity to strangers. I begin by considering general characteristics of Hume’s theory that underpin these results. Moral judgment of an agent’s action at t reflects the sentiment the agent would have by sympathizing at t with the pleasures and pains that result from his action. Recall that the agent will sympathize not only with others but also with his own future pleasures and pains (§3). Since sympathy decays with psychological distance, in the moral point of view we sympathize most intensely with the near-term pleasures and pains of the agent’s narrow circle; the interests of the agent himself, together with his family, friends, and acquaintances, receive special weight. The agent presumably stands in a closer psychological relation to himself than to the other members of his circle. For this reason, “our strongest attention is confin’d to ourselves; our next is extended to our relations and acquaintance; and ’tis only the weakest which reaches to strangers and indifferent persons” (T 488). The agent sympathizes most intensely with himself, so that sympathy with his own future pleasures and pains will result in more intense experiences than sympathy with the feelings of others. This is true holding equal such other factors as the intensity of the future pleasures and pains and their relative proximity in time (cf. 536, 539). We have seen that Hume’s theory is not egoism in its pure or traditional form (§3). It will typically issue moral evaluations that diverge from those of egoism; the pleasures and pains of others (as sympathetically communicated to the agent) are of moral relevance. At the same time, both egoism and Hume’s agent-centered sentimentalism give special weight to the agent’s interests. Hume’s position has a bias in the direction of egoist implications.49 As Hume writes: “our first and most natural sentiment of morals is founded on the nature of our passions, and gives the preference to ourselves and friends, above strangers” (T 491). The extent of this preference depends on empirical facts about the operation of empathy. The greater the difference in intensity between the agent’s sympathy with himself and with his family and friends, the greater the bias toward egoism. By the same token, the greater the difference in intensity between the agent’s sympathy with his (extended) family and with others, the greater the bias toward “tribalism”; the greater the difference in intensity between the agent’s sympathy with his countrymen and with others, the greater the bias toward “nationalism,” and so

49. John Cottingham writes, with Hume’s emphasis on corrections in view, that in his theory “self-preference . . . risks dropping out of the sphere of morality altogether” (1991, 808). To the contrary, self-preference takes pride of place.

hume’s agent-centered sentimentalism

237

forth. The bias toward egoism is a special case of giving greater weight to those with whom one has the closest relationships, where one’s closest relationship is with oneself.50 The moral point of view, for Hume, is not an impersonal, perspectiveless point of view that weighs all pleasures and pains equally; rather, it is partial, in that moral assessment of an agent’s action at t gives disproportionate weight to the pleasures and pains of those psychologically closest to the agent at t.51 There is a considerable literature, from a variety of theoretical perspectives, seeking recognition of a greater role for partiality in ethics.52 Partiality is integral to the moral point of view as Hume conceives it, given that we sympathize more intensely with those with whom we have close relationships. Agent-centered sentimentalism thus provides a possible foundation for partiality in ethics. Hume’s account of the moral point of view generates an agent-centered moral theory.53 Within agent-neutral theories, the fundamental evaluations—whether of the duty-making features of actions or of the valuemaking features of states of affairs—are independent of the agent’s relations to the actions or state of affairs. The mere fact that an act is the agent’s, or that a state of affairs results from the agent’s actions, carries no evaluative weight; value and obligation depend on impersonal or impartial considerations. Utilitarianism is an agent-neutral theory; the values that it assigns to an action’s consequences are independent of the agent’s relation to those states of affairs. Theories that are not agent-neutral are agentcentered. Egoism is agent-centered, since it holds that an agent ought to

50. In his 1953 Herbert Spencer Lecture, Broad outlines a theory he dubs “self-referential altruism,” an altruism limited in scope by obligations founded directly upon one’s special relationships. (Though he does not endorse this position, Broad thinks it approximates that of commonsense Western morality.) See 1953, 277–82. Since Broad allows that we stand in special relations to ourselves (cf. 265), the position could just as well be called “other-referential egoism,” an egoism tempered by obligations founded directly upon one’s special relationships to others. Whatever the label, the agent-centered solution to the problem of variability leads to this position. Stephen Darwall brought the connection to Broad to my attention. 51. Thus I think it quite wrong to interpret Hume as requiring impartiality in the general point of view. See the references at note 15. Also cf. MacNabb 1951/1966, 193; Harrison 1976, 114; Mackie 1980, 67; and Bricke 1996, 107, 139. Similarly, Darwall takes Hume to hold that moral judgments are made from an “impersonal standpoint” and characterized by “agent neutrality” (1999, 141). For previous misgivings similar to my own, see Brown 1994, 26. 52. For references to this literature, see Velleman 1999, 341–42n.10. Among those making the case for partiality, the normative position advanced by Oldenquist is perhaps most similar to that of Hume. Oldenquist emphasizes the prima facie obligations (1982, 182, 187) due to the members of such (typically nested) groups as oneself (cf. 180), family, friends, community members, and countrymen. Oldenquist recognizes that the resulting theory is “neither egoism nor impersonal morality” (176). The theory is a version of Broad’s “self-referential altruism.” See note 50. 53. The characterizations in this paragraph derive from Darwall 1986, 295–99. Oldenquist calls moral judgment “self-dependent” (1982, 175, 180), that is, self- or agent-centered in the sense discussed later.

238

reflection and the stability of belief

act to bring about his own happiness. Deontological theories are agentcentered insofar as they include principles that recognize special duties, or more stringent duties, toward persons with whom the agent has special relationships. In Hume’s agent-centered sentimentalism, the mere fact that a pleasure or pain bears some special relationship to the agent (for example, because it will be his, or that of a member of his family, or of his friend) does carry special moral weight—were it not to bear such a special relation to him, sympathy’s effect on the agent would be less intense. As a result, Hume’s theory has an affinity with egoism, but also with standard deontological theories, theories that recognize special obligations to one’s family and friends.54 In recent literature, there has been considerable discussion of the possibility of agent-centered restrictions within the context of utilitarian theories. Such constraints proscribe an action, even though performing the action would have the best overall consequences, impersonally or impartially conceived. Utilitarianism might want to admit such restrictions, in order to give agents’ personal integrity and agents’ basic projects their proper due. Yet, the problem of providing a rationale for such restrictions within a utilitarian framework has proven quite recalcitrant.55 Might Hume’s agent-centered sentimentalism deliver moral judgments whose content approximates that of agent-centered restrictions? Consider a typical example: we suppose torturing one person would prevent torture to two or three other persons but want to prohibit torture even in these circumstances. It is plausible that Hume’s theory delivers a special case of such a restriction: the theory could prohibit an agent’s torturing his friend in order to prevent torture to two or three strangers; the interests of the friend receive special consideration. In this sort of case, at least, Hume’s theory has the same normative implication as does an agent-centered restriction. The restriction on the agent arises out of the asymmetry in his relationships to the persons affected by his actions. Does Hume’s position nevertheless require the agent to torture one friend (or stranger) to spare two other friends (or strangers), or to torture two strangers to spare a friend? Not necessarily; participating in such torture might be abhorrent to the agent; in sympathizing with the effects of his actions, he will give his own pains special weight, so that they will count for more than in the usual utilitarian calculus. Hume has an additional resource. Suppose the agent is contemplating torturing a person and thus considering producing effects on him. Causation is a relation that strengthens the operation of sympathy (§1). It would seem that a prospective causal relation would also enhance sympathy, intensifying the agent’s sympathetic reaction to the person’s prospective pain.56

54. I have in mind, paradigmatically, Ross’ position in 1930, 19. 55. See Darwall 1986; Scheffler 1982/1994, esp. 24–25 and chs. 4–5; and Kagan 1989, esp. chs. 1–4. 56. This point is due to David Velleman.

hume’s agent-centered sentimentalism

239

Hume’s theory has considerable capacity to yield prohibitions in the neighborhood of agent-centered restrictions.57 Agent-centered restrictions, in the form discussed in recent literature in moral philosophy, are well removed from Hume’s immediate concerns. At the same time, Hume maintains that we have special moral obligations to those with whom we have special relationships. He also maintains that there is a moral requirement of only limited generosity. The agent-centered solution to the problem of variability in sentiment holds promise of enabling Hume to explain both these facts about (what he takes to be) the content of our moral judgment. The theme of partiality first emerges in III.ii.1: [S]ince no action can be laudable or blameable, without some motives or impelling passions, distinct from the sense of morals, these distinct passions must have a great influence on that sense. . . . [W]e always consider the natural and usual force of the passions, when we determine concerning vice and virtue; and if the passions depart very much from common measures on either side, they are always disapprov’d as vicious. A man naturally loves his children better than his nephews, his nephews better than his cousins, his cousins better than strangers, where every thing else is equal. Hence arise our common measures of duty, in preferring the one to the other. Our sense of duty always follows the common and natural course of our passions. (T 483–84) It might seem that Hume is here deriving the existence of special obligations from a thesis about moral motivation. I do not want to deny some linkage. At the same time, the claim that moral obligation requires a sentiment “capable of producing the action” (518) seems too weak, or perhaps too amorphous, to support such a derivation. Let us see what else Hume has on offer.

57. It should not be surprising that Hume’s position is congenial to the content of agentcentered restrictions. Consider Scheffler’s contention that utilitarianism can accommodate agent-centered permissions, though not agent-centered restrictions. Such a permission might give an agent leave not to torture his friend, even though torturing the friend brings about the best overall consequences, impersonally considered. Scheffler’s justification for agent-centered permissions is that people’s motivations typically arise from their own personal point of view and that a moral theory ought sufficiently to reflect this fact. (Cf. his statement of the “liberation strategy,” 1982/1994, 61–62. For an extended discussion of related attempts to generate agent-centered options, see Kagan 1989, chs. 6–9.) It is worth noting the role of motivation in this argument. Scheffler’s permissions allow the agent more room than does utilitarianism to act on his personal motivations. Hume’s sentimentalism roots moral obligation in the availability of motivation in the form of sentiments; his agent-centered sentimentalism roots motivation in sentiments that arise from the agent’s perspective. The result is that some of Scheffler’s permissions (for example, not to torture in specified circumstances) emerge as restrictions. Agent-centered obligations flow directly from Hume’s account of the steady and general point of view.

240

reflection and the stability of belief

A III.ii.2 passage previously quoted leads to an explicit discussion of partiality: [O]ur strongest attention is confin’d to ourselves; our next is extended to our relations and acquaintance; and ’tis only the weakest which reaches to strangers and indifferent persons. This partiality, then, and unequal affection, must not only have an influence on our behavior and conduct in society, but even on our ideas of vice and virtue; so as to make us regard any remarkable transgression of such a degree of partiality, either by too great an enlargement, or contraction of the affections, as vicious and immoral. This we may observe in our common judgments concerning actions, where we blame a person, who either centers all his affections in his family, or is so regardless of them, as, in any opposition of interest, to give the preference to a stranger, or mere chance acquaintance. (T 488–89) Hume’s point is not simply that humans are partial in their conduct, but that they ought to be. The facts about partiality in conduct are reflected in moral judgments—“our ideas of vice and virtue”—themselves. The moral point of view approves, and even requires, partiality. Hume returns to these themes in III.iii.3, “Of goodness and benevolence”: “the generosity of men is very limited, and . . . it seldom extends beyond their friends and family, or, at most, beyond their native country. Being thus acquainted with the nature of man, we expect not any impossibilities from him” (602). In writing of what “we expect,” Hume again has in view moral judgments, not merely predictions about human behavior. The Treatise 602 passage (§1) advancing the stated doctrine of the narrow circle arises in this context: it is “by a sympathy with the sentiments of those, who have a more particular connexion with” any person, that we “form a judgement of his moral character” and “approve of his character.” Hume is describing moral approval, from within the steady and general point of view. As in III.ii.2, the fact that generosity beyond the narrow circle—generosity toward strangers or “indifferent” persons— is limited is reflected in moral judgment itself. Hume’s moral theory is broadly consequentialist and hedonist, focused as it is on pleasures and pains. In the passages we are considering, Hume is prospecting in two areas that pose notorious difficulties for utilitarian theories. The first is doing justice to the obligations generated by special relationships. The second is avoiding the counterintuitive consequence that enormous self-sacrifice is required for the sake of less well-off strangers. Hume seems to proceed, however, by taking note of the facts of limited generosity to strangers and of partiality and then simply announcing that our moral judgments reflect these facts. In III.ii.2, partiality in conduct “influence[s] . . . our ideas of vice and virtue.” Why is that? In III.iii.3, “Being thus acquainted with the nature of man, we expect not any impossibilities from him.” Why not? Hume’s procedure can easily appear ad hoc. Granting the fact of partiality in conduct, why do we lower moral requirements toward strangers

hume’s agent-centered sentimentalism

241

and the whole of mankind? One could fill in an answer, maintaining that ought implies can, or that more demanding expectations are counterproductive, or that special relationships generate special duties. In both III.ii.2 and III.iii.3, however, Hume simply observes that the facts about behavior affect our moral judgments. In III.iii.3, he takes the further step of inferring that we sympathize with the narrow circle; were this not the case, our moral judgments would not recognize special obligations to those within our narrow circle.58 If this is the entire story, it is a brute fact that morality dictates partiality to those with whom we have special relationships, and our sympathizing with the narrow circle is the only answer to the question of with whom we sympathize that is compatible with that fact. Hume appears to take special obligations and limited benevolence as starting points that dictate a specification of sympathy’s target. We can see Hume as doing better if we interpret him in terms of the agent-centered solution to the problem of variability. The material about generosity is located at the second paragraph of III.iii.3, the third context in which Hume addresses the problem of variation in sentiment. His discussion of the question of with whom we sympathize, whose interests we take into account—“We are quickly oblig’d to forget our own interest” (T 602)—merges into a discussion of the perspective from which we sympathize. Hume devotes the second half of the paragraph at Treatise 602–3 to the latter question. The question of sympathy’s target and that of sympathy’s perspective are interdependent (§2), and we find Hume intertwining them in the discussion of generosity. Suppose we construe Hume’s Treatise 602 statement that a moral judge sympathizes with the agent’s narrow circle as a proxy for the claim that a judge sympathizes from within the narrow circle. This intended doctrine of the narrow circle requires further resolution in order to eliminate variability in sentiment. It is at Treatise 591 that Hume poses the question of sympathy’s perspective and broaches the point of view of the agent, of “the person himself.” Since the questions of the distance from which we sympathize and of with whom we sympathize are interconnected, this focus on the agent himself has an egoist feel, and Hume cannot himself stay with it (§3). Suppose, however, that we import the suggestion at Treatise 591 into the discussion of generosity at Treatise 602–3. A direct explanation of the partiality in benevolence is at hand. The variation in sentiment arising from sympathy motivates the adoption of a standard psychological distance. The salient standard is the point of view of the agent. Since we sympathize more intensely with those with whom we stand in close relations, the agent’s point of view is one that sympathizes most intensely with the narrow circle. The interests, the pleasures and pains, of family, friends, and acquaintances carry the greatest weight; “indifferent” persons count for little. Special relationships thus impose moral demands that are absent in connection with strangers. In Hume’s agent-centered

58. Darwall characterizes the direction of the argument in this way (1995, 81n.37).

242

reflection and the stability of belief

sentimentalism, special obligations and a requirement of but limited generosity to strangers are two sides of the same coin. The explanation of partiality in benevolence would be a nifty one, should it succeed. One difficulty is this. The agent-centered theory explains why the moral point of view embodies partiality to those at a close psychological distance from the agent, for they are the persons with whom the agent will sympathize most intensely. Who are these persons? The persons in the agent’s narrow circle are those with whom he has commerce, perhaps “immediate” (T 603) commerce. Hume holds that we are partial to family, friends, and acquaintances. These persons, at least, are members of the agent’s narrow circle. What of rivals, adversaries, and enemies? It is difficult to see how Hume can exclude antagonists from the agent’s “sphere,” the “circle . . . in which [the agent] moves.” In many cases, a rival, adversary, or enemy is a stable part of the scene. But does morality require consideration for one’s friends and rivals equal to that of other members of one’s narrow circle? Hume maintains that special resemblances and relations facilitate and intensify sympathy. He writes that “any peculiar similarity” (T 318) has this effect. It seems natural to suppose that dissimilarities inhibit and weaken sympathy. Sympathy conveys a “conception of our own person” (317) or “consciousness of our own person” (318) to the idea of sentiments of others. An agent and his rival will have dissimilar, even antithetical, goals and concerns, interfering with sympathy’s operation. The point is not that such dissimilarities somehow knock the antagonists outside the narrow circle; the point is that, other relations being equal, we sympathize less with rivals, adversaries, and enemies than with friends. Or so Hume would need to claim. There is the support of empirical evidence that similarities in preferences, attitudes, and interests strengthen empathic response.59 Hume can thus exempt antagonists from the degree of partiality due to other members of the narrow circle. In sum, agent-centered sentimentalism has considerable theoretical power. It generates requirements of partiality to those with whom we have special relationships, on the one hand, and of limited generosity to strangers, on the other. These normative results are derived from an agent-centered conception of the steady and general point of view, where this conception is itself the salient solution to the problem of variability in sentiment due to sympathy. That morality dictates both partiality and limited generosity are therefore consequences of an independent solution to the problem of variability. The explanatory power of the agent-centered solution should, from Hume’s

59. For a summary of relevant studies, see Hoffman 2000, 208–9. There is also evidence that such factors as dislike for another person, or believing that another person is malevolent, an enemy, or in competition with oneself, induces a counterempathic response, pain in response to the other person’s pleasure, and vice versa. See, for example, Zillman and Cantor 1977 and Lanzetta and Englis 1989. It is worth considering whether this phenomenon supports appropriate normative results within Hume’s framework. And there is the prior question whether Hume has at hand associationist resources to explain counterempathy.

hume’s agent-centered sentimentalism

243

perspective, offer some confirmation for his sentimentalism. Unfortunately, his failure to track the distinction between the question of sympathy’s target and that of sympathy’s perspective prevents Hume from self-consciously adopting the agent-centered solution and from laying claim to these theoretical advantages. My goal has been to bring to light the structure of the agent-centered theory, the motivation for it, and some of its more intriguing normative implications. I have also sought to defend Hume’s agent-centered sentimentalism—though not sentimentalism itself—against some objections.60

60. I am indebted to Stephen Darwall and David Velleman for invaluable discussion and comments, and to Justin Broackes, who suggested improvements and saved me from some definite mistakes. I also want to thank a number of philosophers and psychologists who generously responded to “out of the blue” inquiries about the psychological literature on empathy: Dan Batson, Jack Dovidio, John Draeger, Elaine Hatfield, Martin Hoffman, and Stephanie Preston. I am grateful to Phoebe Ellsworth for facilitating these exchanges.

244

reflection and the stability of belief

9 What Is Worth Preserving in the Kemp Smith Interpretation of Hume?

naturalism versus skepticism. Feeling versus reason. These dichotomies, descending from Norman Kemp Smith, have framed the terms of debate in the study of Hume for sixty-plus years. Almost every significant piece of work on Hume has Kemp Smith in the rear-view mirror. Yet, questions that go to the core of the interpretation abound. Is Hume’s naturalism grounded in feeling? Is Humean reason allied with skepticism? Did Kemp Smith put his best foot forward on these matters? Did he even have a consistent view of them? In his inaugural lecture at the University of Edinburgh, Kemp Smith wrote: “Though philosophical systems vary indefinitely, they are reducible, broadly considered, to three main types. They are either idealist, or naturalistic, or sceptical” (1919, 3; cf. 1912, 706).1 Kemp Smith’s 1905 articles in Mind, “The Naturalism of Hume (I) and (II),” established that Hume belongs 1. Page references to Kemp Smith’s publications are incorporated parenthetically within the text, by date of first publication: Studies in the Cartesian Philosophy (1902); “The Naturalism of Hume (I)” and “The Naturalism of Hume (II)” (1905); “Subjectivism and Realism in Modern Philosophy” (1908); “How Far Is Agreement Possible in Philosophy?” (1912); “Kant’s Relation to Hume and Leibniz” (1915); A Commentary to Kant’s Critique of Pure Reason (1918); “The Present Situation in Philosophy” (1919); Prolegomena to an Idealist Theory of Knowledge (1924); ed., David Hume, Dialogues concerning Natural Religion (1935); The Philosophy of David Hume (1941); Lawrence F. Barmann, ed., The Letters of Baron Friedrich von Hügel and Professor Norman Kemp Smith (1981). [Full bibliographical details for these sources may be found in the bibliography in this volume.] I use ‘1905’ to refer to both articles in Mind, which can be distinguished by their page ranges. The 1915 article reappears as xxv–xxxiii of 1918.

245

to the naturalist, rather than the skeptical, type of philosophy. Kemp Smith’s 1941 book, The Philosophy of David Hume, covered much additional ground, but also advanced rich, new interpretive theses. Kemp Smith cites work preparing his 1935 edition of Hume’s Dialogues as the catalyst for the changes (1941, vi). During the decades following the articles, he had been busy with much else. Kemp Smith’s commentary on Kant’s Critique of Pure Reason appeared in 1918, his translation of that work a decade later. He took the occasion of the 1919 inaugural lecture, “The Present Situation in Philosophy,” “to give some indication of my personal sympathies and convictions” (1919, 3). Kemp Smith wrote that he was “speaking as a convinced idealist” (1919, 6). In 1924, he published his Prolegomena to an Idealist Theory of Knowledge.2 I will argue that Kemp Smith’s idealist sympathies obscure his most important contribution to our understanding of Hume. This contribution, I contend, is difficult to reconcile with the conception of the Kemp Smith interpretation that has dominated the Hume literature.

1. The Canonical Understanding of the Kemp Smith Interpretation We need a framework for initial discussion. In the preface to the book, Kemp Smith writes of Hume’s “non-skeptical, realist teaching” (1941, v). His interpretation has spawned extensive bodies of literature elaborating and refining these two key themes, that Hume advances a nonskeptical epistemology and that he is a realist (or at least allows for the possibility of realism) in metaphysics. In order to find an analysis of comparable influence, one has to look to the interpretation Kemp Smith attacked, that of Beattie and Reid. Main elements in the Kemp Smith tradition of interpretation are familiar. There are two assumptions about Hume’s moral theory and its relationship to his epistemology: that for Hume morality is solely a matter of feeling or sentiment and that Hume’s ethics determines his theories of belief and knowledge. It follows that Hume holds a sentimentalist theory of belief as well as morality, that belief and reason are themselves completely subordinate to passion. Then in turn belief is arational or nonrational, and hence impervious to skeptical attack—nonrational because it is solely a matter of feeling or sentiment. Some beliefs nevertheless enjoy nature’s sanction because they are irresistible or inevitable. The emphasis on irresistibility and inevitability is also often supposed to follow from the passional nature of 2. Prichard refers to Kemp Smith’s Prolegomena in four pieces collected in his 1950. He writes: “The idea that perceiving is a species of knowing has recently become prominent in an alleged refutation of Berkeley, which has not infrequently been advanced of late years, and notably by Professor Moore, Mr. Bertrand Russell, and Professor Kemp Smith” (202); he proceeds to quote and discuss Kemp Smith at some length. The 1905 articles, however, do not seem to have influenced Prichard’s lectures on Hume.

246

reflection and the stability of belief

belief. The claims that morality is a matter of feeling and that the ethics determines the theory of Book I more or less dictate the subordination of belief and reason, the nonrationality of belief, and the doctrine that nature sanctions irresistible beliefs. Finally, there is a claim about the content or scope of the natural beliefs. Whereas Hume emerges as “nonskeptical,” courtesy of the claim that nature sanctions some beliefs, he is “realist,” courtesy of the claim that the beliefs in body and causal necessity are natural beliefs. The Kemp Smith tradition offers a neat interpretive package. I intend this summary as a canonical or stereotypical representation of the Kemp Smith interpretation, extracted from literature on his work.3 We will never understand Hume if we insist, as does Kemp Smith as standardly portrayed, on the complete subordination of belief and reason to passion and on the nonrationality of belief.4 In the articles, Kemp Smith’s detailed exposition of Hume’s theory of belief puts surprisingly little weight on the thesis that belief is a matter of sentiment. Following Hume’s exposition in both the Treatise and the first Enquiry, Kemp Smith discusses belief in two stages. First, there is Hume’s view that inductive belief is due to custom. Kemp Smith interposes a revealing comment: “It would, Hume remarks, be quite allowable to stop our researches at this point, taking custom as a natural propensity of the soul conditioning belief; but, as it happens, we can carry our inquiries a step further” (1905, 163). The further step, the second stage, is Hume’s view that belief is a feeling that results from enlivening. According to Kemp Smith, for Hume it would be “quite allowable” were the role of custom to close out the theory, though “as it happens” we can identify a role for feeling. This language, repeated verbatim in the book (1941, 122), is quite extraordinary if Kemp Smith considers feeling to lie at the core of Hume’s theory of belief and knowledge, and to generate the subordination of belief to passion and the nonrationality of belief.

2. Kemp Smith on Hume’s Reconstruction of Reason and of Normative Discriminations A conception of belief as passional and hence nonrational indeed conflicts with a central feature of Kemp Smith’s interpretation. He writes: “my chief aim will be to state the grounds of [Hume’s] naturalistic view of reason, and to show how his philosophy of knowledge culminates in a new theory of belief” (1905, 158; cf. 1905, 335). I wish to focus not on the “new theory of belief,” but on the “naturalistic view of reason.” 3. Sources for this understanding of Kemp Smith include Beauchamp and Rosenberg 1981, 57–59 (but see note 27 here); Norton 1982, esp. 5–6, 16–18, 56; and Garrett 2005, xxv–xl, as well as early reviews—Ewing 1942 and Jessop 1948. 4. Literature (beyond sources cited in other notes) critical of various aspects of the Kemp Smith interpretation includes Hendel 1963; Mijuskovic 1971; Stove 1977, §1; Fogelin 1985; J. Moore 1995; Mounce 1999, ch. 1; and Wright 2007.

the kemp smith interpretation of hume

247

In order to locate the conflict, we need to take into account Kemp Smith’s distinction between “analytic” and “synthetic” reason (1905, 156–57; 1941, 99–100). Analytic reason is “insight” (1905, 151, 165; cf. 1941, 44, 65), “rational insight” (1905, 155; cf. 1941, 102), “purely intellectual” (1905, 165), a faculty that uncovers “logical necessity” (1905, 157), “discovering necessary relations between ideas” (1905, 342). Synthetic reason “embraces all knowledge outside mathematics” (1905, 157). Kemp Smith intends this distinction to track that in Hume between “demonstrative reasoning” concerning relations of ideas and “moral” (EHU 35) or “experimental” (108, 165) reasoning concerning matters of fact and existence.5 His claim is that Hume provides a naturalistic reconstruction of synthetic reason. Kemp Smith expresses this idea in two different ways. First, Kemp Smith takes Hume to identify reason with instinct, the source of the natural beliefs: “reason in [the] synthetic sense . . . [is] itself fundamentally instinctive in character” (1941, 65; cf. 199–200); “Reason, [Hume] roundly declares, is ‘nothing but a wonderful and unintelligible instinct in our souls’” (1905, 157; 1941, 100—cf. 199). Second, he takes Hume to identify reason with the natural beliefs themselves: Hume uses ‘reason’ “sometimes as a title for the natural beliefs” (1941, 461; cf. 64–5); “Reason . . . is nothing distinct from our natural beliefs” (1905, 156; cf. 166). The problem is that it is odd to contend both that Hume holds a naturalistic view of what constitutes reason and that reason is nonrational. It would seem that if “reason” is fundamentally nonrational, then Hume does not advance a naturalistic conception of reason at all. One might try to explain away the conflict as an artifact of Kemp Smith’s terminology. Kemp Smith writes of various beliefs and inferences that they “cannot be justified by reason” (1905, 151, 154; 1941, 86), that they result from “some unreasoning propensity” (1905, 163; 1941, 121), from “nonrational” principles (1905, 157, 168; 1941, 100, 454; also 1915, 289, and 1918, 594). He also writes that they cannot be “theoretically justified” (1905, 162; 1941, 116). Perhaps Kemp Smith means that these beliefs and inferences are nonrational in that they are not due to analytic reason. The book, however, cannot always be read in this way. Kemp Smith recognizes the conflict, asking, “Can reason be non-rational?” (1941, 64). He proceeds: “Reason, in the sense of analytic reason, certainly cannot be described as non-rational: but if reason be also employed as a name for certain ultimate beliefs . . . , what is there to prevent ‘reason’—reason in this synthetic sense— from being itself fundamentally instinctive in character?” (1941, 64–65). This is not the triviality that synthetic reason is not analytic reason; the point is that synthetic reason, unlike analytic reason, can be and is described as nonrational. Kemp Smith takes Hume both to identify synthetic

5. [In this chapter, quotations of the Treatise and Enquiries are based on the volumes edited by Norton and Norton (NN) and Beauchamp (B-98, B-00) in the Clarendon edition of Hume’s works.]

248

reflection and the stability of belief

reason with instinct or natural belief and to hold that instinct is deeply nonrational. Another hypothesis is that Kemp Smith misrepresented his position. Perhaps he did not intend that Hume identified synthetic reason with instinct or natural beliefs. Perhaps he thought the point of Hume’s naturalism is to show that there is no such thing as synthetic reason, much as the point of eliminative materialism is to show that there are no such things as minds. If so, Kemp Smith read Hume as maintaining that we mistake nonrational instincts or beliefs for synthetic reason. Kemp Smith, however, contends that he is faithfully representing Hume’s own terminology, that ‘reason’ is employed by Hume as a name for certain ultimate beliefs or instincts. The distinction between analytic and synthetic reason captures “the two very distinct meanings which [Hume] ascribes to the term ‘reason’” (1905, 156; cf. 1941, 99). Similarly, after noting that synthetic reason “is much the more important” (1905, 157; cf. 1941, 65), Kemp Smith observes that “Hume constantly equates it with reason in general” (1905, 157; cf. 1935, 36n.3, and 1941, 100). Kemp Smith is clear that Hume’s terminology is no mere courtesy.6 In the book, “with the sole exception of ‘analytic reason’, all the cognitive faculties” are founded on the imagination (1941, 219); “Hume speaks as if the ordinary conceptions of knowledge will remain unaffected, and will find in human nature a more secure foundation than any hitherto provided” (1941, 63).7 This statement, within a paragraph discussing Hume’s claim that “reason is nothing but a wonderful and unintelligible instinct” (T 179), reveals the depth of Kemp Smith’s commitment to the thesis that Hume naturalizes synthetic reason and knowledge. This does not sit well with the claim in the stereotype that belief is nonrational. The book elaborates Hume’s naturalistic reconstruction of reason in ways that exacerbate the tension. A new distinction, a two-tier structure between “fundamental” and “derivative” or “specific” beliefs, sets the stage. As a preliminary step, the natural beliefs are reduced to two. In the articles, the natural beliefs include belief in body (1905, 151–52), causal action (152–53; cf. 161– 62), the self (153–54), and material substratum (159–60; cf. 170–71); in the book, the belief in objects with a continued existence, and the belief in causal connection (1941, 116, 124, 222, 409–10, 455, 483, 486, 503, 548–49)—“continuants” that are “causally active” (1941, 543). These are the “fundamental” (1941, 45, 46, 68, 388) beliefs, “those . . . which decide our view of the general character of the environment, physical and social, in which our lives have to be lived—predetermined, independently of all special individual experience” (1941, 46). The fundamental beliefs “precondition—on a deeper, more constant level, as it were—all our innumerable, more specific beliefs” (1941, 124; cf. 222). These latter are the “derivative beliefs to which our specific

6. Laird was another early commentator to emphasize this feature of the Treatise (1932, 104–6). 7. But see the context, cited later (in §4), of the second of these passages.

the kemp smith interpretation of hume

249

experiences give rise” (1941, 388; cf. 455–57).8 The fundamental beliefs are those “other than [the] merely custom-bred” (1941, 83). By contrast, the specific beliefs are acquired “through causal ‘inference’” (1941, 124) and hence through custom, with Hume thus “equating . . . experimental reasoning with custom-bred expectation” (1941, 64).9 Empirical reasoning receives much more attention in the book. Accordingly, there emerges a new interpretive slogan, “custom is king” (1941, 95), that is absent in the 1905 articles. The book’s new focus on the specific, or custom-bred, beliefs involves two main elements. In the first place, Kemp Smith calls attention to a causal theory of assurance in Hume. He thus devotes a number of pages to the I.iii.9 discussion of two systems of realities (T 107–8)—one based on the senses and memory alone, the other connected to the first by custom, by the relation of cause and effect (1941, 383–85; cf. 236). This material is one of Hume’s chief expressions of his claim that cause and effect is the only relation that allows us to advance beyond perception and memory. In passages Kemp Smith cites, cause and effect is the one relation that “informs us” (T 74) of, and “brings us acquainted” (108) with, objects we have not perceived (1941, 366, 368, 384). There are many similar passages in Hume (T 73, 103, 104; Abs. 654; EHU 55, 164). Kemp Smith writes that “Hume’s ultimate criterion of truth is conformity with the realities of these two systems” (1941, 385), faithfully reflecting Hume’s own tone of special approval of causal inference.10 In the second place, Kemp Smith emphasizes that Hume draws normative distinctions within custom: “Hume himself concedes that not all customs are good customs” (1941, 383). The bad customs include the beliefs of liars who come to believe their lies and beliefs based on “education,” indoctrination (1941, 378, 383, 427, 460–61). These beliefs result from repetition, the repetition of an idea, and hence from custom, but not specifically from the relation of cause and effect. Further, Kemp Smith recognizes that even beliefs based on observed conjunctions do not win Hume’s approval in every case— not if the regularities are not “truly causal,” resting instead on “contingently determined combinations of causes” (1941, 385–86); not if the “regularities hold only for a limited experience” (1941, 383); not if “customs conflict” (1941, 94; cf. 383); and not if the “recency” of an experience distorts the

8. Kemp Smith has some tendency to restrict the natural beliefs and synthetic reason to the fundamental beliefs (cf. 1941, 46, 65), but also some tendency to allow that the specific beliefs as well are natural beliefs and due to synthetic reason (cf. 1941, 388). It would be odd if custom-bred “empirical reasoning” were not included in “synthetic reason.” The more inclusive applications of “natural belief” and “synthetic reason” are congenial in contexts where Kemp Smith wants to emphasize that both the fundamental and specific beliefs are ultimately due to instinct, the more restricted applications in contexts where he wants to emphasize that only the more specific beliefs are custom-bred. The former contexts predominate, for reasons that emerge in §5. 9. The claim that “empirical reasoning” is “custom-bred” can also be stitched together from passages in Mind (1905, 158, 164). 10. Kemp Smith was not the first to spot Hume’s epistemic regard for causal inference. See G. E. Moore 1909, esp. 149–51, 154–55, and Price 1940, 31–33.

250

reflection and the stability of belief

results of custom (1941, 387–88). Kemp Smith is thus led to distinguish between the bad customs and “experience,” experience in a “normative sense,” “experience . . . which is, and ought to be, the ultimate court of appeal” (1941, 382; cf. 61). Kemp Smith writes that “Hume’s real position is not that custom (or habit) as such is king” (1941, 382; cf. 386), but that the good customs are king.11 Kemp Smith is attributing a variety of normative discriminations to Hume. Whether or not we dignify these distinctions with the appellation ‘epistemic’, the suggestion that Kemp Smith’s Hume takes all belief about matters of fact to be nonrational grossly misrepresents his interpretation. Beliefs based on perception, memory, and the relation of cause and effect— beliefs comprised in the two systems of realities—have a preferred standing. Not all memories or customary inferences are on equal footing. The book thereby provides further reason to insist that Kemp Smith takes Hume to provide a reconstruction of empirical reason—identifying empirical reason with the good customs within causal inference. Kemp Smith’s Hume is far from providing a debunking explanation of all synthetic reason.

3. Accounts in Kemp Smith of the Basis for Hume’s Normative Discriminations We can encapsulate the advance of the book in the following thesis: Hume identifies good empirical reasoning with appropriately refined or processed products of customary transitions from causes and effects, and thus draws normative discriminations among beliefs about matters of fact. What are the conditions under which beliefs or belief-forming mechanisms enjoy nature’s “sanctions” or “authority” (cf. 1941, 454, 486, 564), a positive epistemic status? My discussion of this question will be brief, just sufficient to take note of the kinds of options Kemp Smith considers. One complication deserves mention. Kemp Smith needs to identify a Humean epistemology that will account for the favorable status of fundamental and specific beliefs alike. The fundamental beliefs in continuants and causality are theoretical, rather than moral or practical. Kemp Smith thus construes them as general beliefs in matters of fact or existence (1941, 549). Yet, he is explicit that fundamental beliefs are not custom-bred (§2) and hence not based on the relation of cause and effect. (Kemp Smith is right to say this. The belief in body, at least insofar as it arises via constancy of perceptions, is not due to the relation of cause and effect. In the case of causal necessity, the felt determination of the mind is a by-product of repeated observation of conjunctions, but the projection of that feeling onto objects

11. Beauchamp and Rosenberg (1981, 54) see this strand in Kemp Smith, though not its inconsistency with their primary representation of his views at 57–59. See also Brand 1992, 54–55.

the kemp smith interpretation of hume

251

is a further step due to an imaginative propensity other than custom.) The beliefs in body and necessity constitute exceptions, counterexamples, to a principal interpretive innovation in the book—that good custom is king, that all knowledge of matters of fact is based on perception, memory, and the relation of causation (§2).12 Fortunately, we need not pursue this matter. The sorts of epistemological positions Kemp Smith tends to attribute to Hume have some capacity to apply both to fundamental and specific beliefs. Consider Kemp Smith’s focus on “unavoidability” (cf. 1941, 116–17), “inevitability” (cf. 1905, 152, 159, 160; 1941, 87), “irresistibility” (cf. 1905, 161, 162; 1941, 486)—“psychological compulsiveness” (1941, 46).13 Even specific beliefs might be irresistible, once triggered by special experiences. According to the canonical representation, Kemp Smith takes irresistibility and inevitability, properties that can be understood in purely psychological and hence “naturalistic” terms, to do the epistemological work. Kemp Smith often strikes out in the direction of other naturalistic properties, emphasizing Hume’s reliance on Hutchesonian, biological analogies (1941, 224; cf. 284–85 and 1935, 27). Natural beliefs are “the outcome of the ultimate propensities that constitute our human nature” and thus “can be shown in their perfect fitness to the calls which things make upon us, to be as wonderfully adapted as any of the animal instincts” (1905, 155). Thus, natural beliefs are “reliable and legitimate,” but “only in practical life” (1905, 168; cf. 1915, 289, and 1918, 594–95); “our ultimate and unalterable tendencies to action are the test of practical truth and falsity” (1905, 156; cf. 170–71). In this vein, Kemp Smith writes of “a new conception . . . of the function of reason” (1905, 155). Kemp Smith writes in the book of “customs and habits that are reliable and beneficial” (1941, 382; cf. 95, 383). Similarly, some “processes of mind . . . are adaptive” (1941, 76); various principles or belief-forming mechanisms “are necessary for [the mind’s] proper functioning” (1941, 493–94; cf. 131). These observations are suggestive

12. Kemp Smith could point out that, strictly speaking, Hume’s causal theory of assurance is the claim that causation is the only relation that informs us of matters of fact (1941, 366, 383) and that the fundamental beliefs are known on some basis other than inference from relations. Or he could remind us that the fundamental beliefs are “classed by themselves” (1941, 124) insofar as they precondition specific beliefs. Such maneuvers circumvent the spirit in which Hume offers the causal theory of knowledge. Kemp Smith misses an opportunity fully to pursue the thought that one of Hume’s underlying projects is to reduce all synthetic knowledge to perception, memory, and causal inference. He sees that Hume relies on his causal theory of knowledge in his discussion of judgments of identity (1941, 368). Even in the case of the belief in body arising from the coherence of perceptions, “The inferences made are causal inferences,” albeit arising “from custom only in an indirect and oblique manner” (1941, 472), but without relying on projection. The question arises whether Hume might have taken this line of investigation further. 13. The following are testimony to the enduring influence of Kemp Smith’s idea: Lenz 1958; P. F. Strawson 1958 and 1985, 10, 11; Stroud 1977, 14, 76, 115, 247–48; Ayer 1980, 69–71; F. Wilson 1997, 113–20; and Millican 2002, §§11–12.

252

reflection and the stability of belief

of the interpretive thesis that Hume approves of beliefs that result from mechanisms that are adaptive or from the proper functioning of belief-forming mechanisms.14 Kemp Smith’s work points in the direction of a class of interpretations of Hume’s epistemology. The crucial notions in these interpretations— whether psychological (irresistibility, inevitability) or biological (proper functioning, adaptiveness)—are all broadly “naturalistic.” In each case, epistemic justification—“Nature’s sanctions”—supervenes on one naturalistic property or another. We can thus generalize the interpretive doctrine of natural belief to the claim that, for Hume, beliefs are justified in virtue of naturalistic properties, either of the beliefs themselves or of the mechanisms that produce them. A naturalistic epistemology is the essence of Hume’s “naturalism.”15 Every account of Kemp Smith takes note of some version of the claim that belief is “natural.” Kemp Smith goes beyond this: his Hume provides “a naturalistic view of reason” (§2). This feature is almost always neglected in discussion of the Kemp Smith interpretation. Why is this so? One explanation is that the 1905 articles scarcely mention normative discriminations among beliefs and give no play to the causal theory of knowledge or to the distinction between good and bad customs (§2).16 Even in the book, the only sustained discussion of these matters occupies but a dozen pages (1941, 376–88).17 There is another, more subtle, obstacle to recognition of the role of the naturalistic reconstruction of reason in Kemp Smith. Though appreciative of the textual evidence for attributing such a reconstruction to Hume, Kemp Smith is in conflict about this element of his interpretation. It is here that his idealist sympathies get in the way.18

14. For a proper function interpretation of Hume, see Craig 1987, 81, and Wolterstorff 1996, 166n.6; for discussion of the adaptive option, see Schmitt 1992, 68–72. 15. For Hume’s motivation for adopting an epistemology in this region, and some elaboration on the options available to him, see my 2006 [this volume, ch. 10] and 2007 [this volume, ch. 12], esp. §§1–4. 16. The articles contain none of the book’s examples of beliefs arising from custom that Hume rejects (§2). Kemp Smith does write in Mind: “The understanding is nothing but the imagination acting according to its most general and established habits or instincts” (1905, 166). In a footnote, he cites Hume’s distinction (T 225) between the irresistible, permanent, and universal principles of the imagination and those that are changeable, weak, and irregular (1905, 166–67n.1). Curiously, in the book there is only one mention of this key passage (1941, 459–60), where it arises somewhat incidentally in an appendix. But see note 37. 17. Ewing wrote: “I think that Kemp Smith ought to have discussed more fully Hume’s not altogether successful attempt to distinguish between the non-rational beliefs which we ought and those we ought not to accept. Obviously any such theory as Hume’s is only defensible if such a distinction can be made within it, otherwise we shall have no more justification for accepting the best thought-out scientific belief than for believing the crudest superstition” (1942, 70). It is not clear from his review that Ewing was aware that Kemp Smith discusses the matter at all. 18. Wright makes a suggestion along these lines (1983, 211).

the kemp smith interpretation of hume

253

4. Distortions due to Kemp Smith’s Idealism Reviews of Kemp Smith’s Prolegomena to an Idealist Epistemology—by Blanshard, Lovejoy, and Mackenzie—all begin by lamenting the title, on the ground that the work presents a “realist” theory of knowledge as a partial foundation for an idealist theory of reality at which the book only gestures.19 In what sense was Kemp Smith himself an idealist? Idealism, naturalism, and skepticism, “the three main types” of “philosophical systems” in the inaugural lecture, constitute “The three possible philosophical attitudes” in a lecture handout (TPPA) carrying that heading.20 Kemp Smith considers Socrates, Plato, Descartes, Kant, and Hegel representative idealists (1919, 5). (TPPA omits Socrates and adds Spinoza.) What does he take these figures to have in common?21 In the Prolegomena, Kemp Smith uses ‘idealism’ “in a very wide sense, as covering all those philosophies which agree in maintaining that spiritual values”—in phraseology repeated in TPPA—“have a determining voice in the ordering of the Universe” (1924, 1). Within idealism, spiritual values include aesthetic, moral, social, and religious values (1919, 6, 16, 20, 25). Further, for idealism, “intellectual and spiritual values stand on the same plane of objectivity, and therefore justify parity of treatment” (1919, 15). In idealism, both spiritual and intellectual values play a determining role in the structure of the universe. “[S]piritual values have been present from the start” (TPPA, though the remark is confined to spiritual values). In the Prolegomena, “The alternative position, . . . ‘naturalism’, is that [spiritual] values emerge, and begin to vindicate their reality, only at some later stage in a process of evolution” (1924, 1). In the inaugural lecture: According to naturalism, parts of the Universe are more complex and are more completely unified than is the Universe as a whole. Certain 19. Blanshard 1925; Lovejoy 1925; and Mackenzie 1924. 20. The “Papers of Professor Norman Kemp Smith (1872–1958),” University of Edinburgh Library, Special Collections and Archives, boxes 1416–1422, include eleven one- or two-page, undated handouts—four on prehistoric and Greek philosophy, two on secondary qualities and the representative theory of perception (central issues in his Prolegomena), and a group consisting of “The Nature of Reasoning and of Proof” (NRP) and untitled handouts on truth (T), the hypothetical-deductive method, and fact and theory, as well as TPPA. The handout on hypotheses also relies on the trichotomy between idealism, naturalism, and skepticism. These five items are included in boxes 1420 and 1421. The handouts are in blue or purple ink, suggestive of mimeograph (ditto) or spirit duplicator (Banda) reproduction. Among other materials in box 1421 are university documents, similarly reproduced, dating from the early 1930s. 21. Maclennan writes: “[Kemp Smith] did not believe that the nature of the part is determined by that of the whole” (1967, 43). I suspect Kemp Smith’s view is more complicated, in light of passages I cite later in this section and 1924, 163–64. Maclennan, I think, is quite right that “[Kemp Smith] rejected the Idealist doctrine of Coherence completely as a definition of truth, though he retained it as a test of truth” (43). Kemp Smith writes: “Truth may be defined as follows: Truth is a reality stated in the form of propositions (or judgments) by means of language or some other body of symbols” (T). There is more on coherence in NRP.

254

reflection and the stability of belief

parts, too, possess higher qualities, such as life and consciousness, which are not to be found in the wider reality that includes them. . . . The universe is, as it were, merely the stage, and is not itself a center of interest.22 (1919, 18) This leads to a criticism: “Naturalism has to treat human values as merely relative” (1919, 18), relative to a part of the universe. The criticism applies, as we should expect, to naturalism’s treatment of intellectual values, but not to the “idealist view that . . . logical criteria have absolute validity, that knowledge is really knowledge, that is to say a form of genuine insight, revealing to us the independent[ly] real” (1919, 14; cf. 10, and 1981, 50, a 1919 letter about the inaugural). In the inaugural lecture, Kemp Smith grants that “man is the most complex existence known to us”; nevertheless, “our specifically human experience” provides only “clues” on which “idealism bases its ultimate conclusions” (1919, 18). He elaborates: “In the view of a naturalistic philosophy, man is a being whose capacities, even in their highest activities, are intelligible only as exercised exclusively in subordination to the specific requirements of his terrestrial environment” (1919, 25); and in the 1919 letter, “my main point” is that “the differentia of naturalism is its attempt to solve the problems of life from a merely planetary & human stand-point” (1981, 50).23 In the Prolegomena, “What is most truly distinctive in idealism is its central contention, that spiritual values can be credited as operating on a more than planetary, that is, on a cosmic scale” (1924, 3–4).24 Naturalism misconstrues a distinctively human outlook, a product of evolution, for the truth about the universe as a whole. Kemp Smith takes the epistemic theories he locates in Hume to be forms of naturalism, par excellence, and hence thoroughly ill-conceived. The fundamental beliefs, the beliefs that are not custom-bred, are “determined by the constitution of our human nature, in an instinctive . . . manner” (1941, 454; cf. 458). The irresistibility and inevitability of these beliefs are due to

22. In TPPA, “Naturalism maintains that life, mind, consciousness, personality, and the values which they condition, first emerge and begin to indicate their reality, at a later stage in the process of evolution. Accordingly, on this view, the Universe has no meaning or value as a whole; there can be meaning or value only in certain of its parts. Indeed . . . as naturalistically conceived it is possible that [the Universe] may turn out to be merely an aggregate.” 23. Garrett is correct that a tenet of Kemp Smith’s idealism is that “values play an ultimate role in the explanation of the structure of the universe” (2005, xxix). Garrett, however, has nothing to say about the thesis that the structure of the universe cannot be discerned by way of distinctively human propensities or capacities. As we shall see (§4), it is this thesis that illuminates Kemp Smith’s treatment of Hume in his book. Sebastian Gardner hits all the right notes in his exposition of Kemp Smith’s philosophical position (2007, 21–27). 24. In TPPA, “while naturalism may be no less insistent than Idealism that the criteria yielded by spiritual values exercise a determining influence in human affairs, what is distinctive of Idealism is its contention that these values can be credited as operating on a more than planetary, on a cosmic scale.” Language similar to the first half of the sentence also appears at 1924, 4.

the kemp smith interpretation of hume

255

instinct (1941, 116–17, 486). The “natural beliefs are . . . reliable and legitimate,” but “only within a strictly limited domain” (1941, 128), “when they are duly proportioned and in keeping with their natural conditions” (1941, 131). Naturalism tends, in the words of the inaugural lecture, to a “pragmatic view of knowledge” (1919, 14). Whereas idealism seeks “absolute validity” and access to the “independently real,” Hume’s naturalism is “incapable of supplying an absolute standard” (1941, 147); belief “allow[s] of no kind of absolute or metaphysical justification” (1941, 407). The derivative beliefs are due to custom, which is itself instinctive (1941, 64, 100), so that its operation and effects are similarly “essential for the maintenance of the individual and the species” (1941, 123; cf. 493–94). All synthetic belief “rests . . . on certain instinctive propensities proper to the human species” (1941, 65; cf. 100, 199–200, and 1905, 163). The fundamental and specific beliefs alike result from instinct and thus are “merely relative,” “relative to ‘the particular fabric and constitution of the human species’” (1941, 147; cf. 65, 1905, 343, and 1935, 31).25 Nature ensures that all synthetic belief serves the distinctive needs of the human species in its environment. This is not to be confused with truth about the universe as a unified, complex whole. Kemp Smith has a regrettable tendency to read his own idealist reservations about naturalism into Hume. Kemp Smith’s accomplishment is to see that Hume takes himself to provide a naturalistic reconstruction of synthetic reason—“equating” reason with custom, thereby placing knowledge on a “more secure foundation”—and drawing normative discriminations within custom, and hence within instinct itself (§2). Yet, Kemp Smith writes: “In the field of matters of fact and existence . . . the term ‘reason’, if still held to, is a name only for certain fundamental beliefs to which we are instinctively and irrevocably committed” (1941, 68, emphasis added). Similarly, he writes of Hume’s account of “the so-called understanding” (1941, 219, emphasis added). Elsewhere, Kemp Smith places both ‘reason’ and ‘understanding’ in scare quotes (1941, 461; cf. 392).26 These passages suggest that Hume is providing a debunking account of synthetic reason after all. Kemp Smith pushes this theme hard. Just as, on this account of Hume, there is no such thing as reason, there is no such thing as synthetic knowledge: “May not our so-called judgments of knowledge in regard to matters of fact and existence be really acts of belief, not of knowledge . . . ? Such is indeed the positive teaching of Book I of the Treatise” (1941, 44); “all judgments of belief express an attitude which does not permit of being equated with any species of knowing or understanding” (1941, 396). There are many similar statements in 25. Kemp Smith amends a 1740 letter from Hume to Hutcheson (1905, 342n.1; 1941, 19–20) where sentiments of approval and disapproval arise “from the particular Constitution of your Nature” (LDH 1.39). Maclennan puts this nicely: “[Kemp Smith] stood for the general human values, but he would have deplored the attempt to establish them upon human nature and human society alone as in the end destructive of those values themselves” (1967, 41). 26. Garrett briefly notes that Kemp Smith sometimes places such terms as ‘inference’ and ‘reason’ in scare quotes (2005, xxxiii), but he overlooks the point that this strand in Kemp Smith is antithetical to the thesis that Hume supplies a naturalistic reconstruction of reason.

256

reflection and the stability of belief

the book (cf. 1941, 222, 400, 446, 447). Kemp Smith’s is not the terminological point that Hume confines ‘knowledge’, strictly speaking, to analytic reason uncovering relations of ideas (T 124). He maintains that association by the relation of cause and effect does not even produce empirical evidence: “When the mind passes from an idea or impression of one object to that of another, it is the imagination which is operating, not the understanding. It is custom and not reason, habit and not evidence, which is at work” (1941, 375; cf. 11, 85, 392); “a natural . . . belief . . . rests neither on insight nor on evidence” (1941, 485–86; cf. 10, 13, 391). Since synthetic belief is not based on evidence of any kind, it constitutes mere opinion, not knowledge in any good sense of the term.27 In a related theme, “There is, [Hume] argues, no such thing as causal inference” (1941, 375). This is an oft-repeated claim in passages where Kemp Smith is discussing Hume’s account of empirical reasoning (1941, 372, 387, 392, 453), the good kind of custom. Inference joins the parade of phenomena with a cognitive flavor demoted by scare quotes (1941, 124, 399, 492).28 Kemp Smith often contrasts causes with both reasons and evidence. He writes that, for Hume, “in all ultimate issues, belief rests on causes only, not on grounds or reasons” (1941, 486; cf. 454, 458). He generalizes the point to the whole of synthetic reason: “all . . . judgments concerning matters of fact and existence . . . [rest] upon feeling, not upon reason, i.e. upon causally determined propensions” (1941, 482). Does Kemp Smith think that causes can never be reasons? His discussions are perhaps guided by a more specific idea, that causes grounded in distinctively human propensities or instincts are not reasons. Causes are not reasons because instincts—whether human or in nonhuman animals (1941, 70, 453, 539)—tailored to the earth’s environment are not reasons. Thus, “our so-called judgements of knowledge in regard to matters of fact” are indeed mere acts of belief, which, “like all passions, [are] fixed and predetermined by the de facto frame and constitution of our human nature” (1941, 43–44). The subordination of belief to “feeling” is at bottom a subordination, one which idealism roundly rejects, to instinctive operations adapted to the terrestrial environment. This result explains Kemp Smith’s relatively favorable attitude toward Hume as represented in a strand in his interpretation that I have not yet mentioned. Kemp Smith writes that Hume relies on reason in a way that results in a “more positive view of the relation of reason to feeling and instinct” (1941, 128–29), more positive than taking reason to consist in instinct alone. Although the same statement appears in the articles (1905, 169), in the book, ‘reason’ is explicitly a label for “reflective thinking” and “[p]hilosphical reflection” (1941, 128). This terminology is symptomatic of a recurring interpretive theme, that it is “only those customs which can 27. Beauchamp and Rosenberg take note of these claims in Kemp Smith (1981, 60–65), commenting: “The problem lies not with Hume. It lies with Kemp Smith” (63). But they do not see that his idealism is the culprit. 28. Jessop’s uptake is on the debunking strand: “What [Hume] is denying is causal inference” (1948, 267).

the kemp smith interpretation of hume

257

survive reflective scrutiny, which ought to be relied upon” (1941, 388; cf. 94–95, 100, 128–29, 422).29 This has a rather different flavor than the emphasis, spanning the articles and the book, on such psychological and biological notions as irresistibility, adaptiveness, and proper functioning (§3). I suspect Kemp Smith conceives of reflection as independent of instinct and the constitution of the species, perhaps even independent of causes altogether, so that its employment does not succumb to his idealist scruples. Since, in his picture of Hume, reflection merely supplements custom and instinct (cf. 1941, 68, 128, 539), the naturalistic epistemology remains wrongheaded. Kemp Smith nevertheless sees Hume as having taken a step in the right direction, in the direction of authentic reasons. The thought that Hume assigns a key role to reflection, as Kemp Smith conceives it, is more congenial to his favored idealism, providing Kemp Smith an avenue for trying to save Hume from naturalism. Ironically, Kemp Smith ought not to have allowed any normative role for reflection to assuage his idealist objection to Hume’s naturalism. Hume is thoroughgoing in his reduction of experimental reasoning to custom and hence to instinct. For Hume, reflective processes are themselves higher-order customs that are more reliable, more adaptive, and so forth (cf. T 149–50 and EHU 107n.). Hume naturalizes synthetic reason “all the way up.” Wearing his idealist blinders, Kemp Smith just cannot fathom (but cf. 1941, 129) that Hume would undertake this project. In the articles, there is comparatively little sign of Kemp Smith’s idealism undercutting his interpretation. Kemp Smith writes once of “supposed inferences” (1905, 155) and ‘rational’ appears once in scare quotes (1905, 166), though the intended contrasts may be with analytic reason (§2). He allows that for Hume there is such a thing as inference: “the ideas introduced by [custom] are, as we say, ‘inferences’, and not mere suggestions” (1905, 163); “Inference . . . is itself identical with [the relation of cause and effect]” (1941, 164; cf. 171). Reason and knowledge are frequently said to be “practical” (1905, 155, 156, 157, 168, 169), but this is not disparaging; Hume is providing “a new conception . . . of the function of reason”—a “naturalistic view of the function of reason” (1918, 595)—“to afford us guidance in practical life” (1905, 155). Most revealingly, “[Hume’s] philosophy is throughout inspired by a new conception of knowledge” and of “[t]he function of knowledge” (1905, 155). This, as well as quotations canvassed in §2, is far from the contention of the book that for Hume there are no such things as inference, reason, and knowledge. Of course, the natural beliefs avowedly have a Kantian resonance (1905, 155). Kemp Smith had studied with idealist teachers.30 “Subjectivism and Realism in Philosophy” (1908) and, to a lesser extent, “How Far Is Agreement Possible in Philosophy?” 29. There is scarcely an interpretation of Hume’s epistemology that Kemp Smith does not foreshadow; in this instance, reflective endorsement interpretations in the fashion of Baier 1991, and Korsgaard 1989 and 1996. 30. See Ewing 1959, 298.

258

reflection and the stability of belief

(1912) were distant precursors to the Prolegomena (cf. 1924, 9). We can hypothesize that Kemp Smith’s own idealism did not gel until the second decade of the twentieth century, in connection with work on his commentary on the first Critique. When Kemp Smith treats synthetic reason as nonrational in the articles (§2), I think he generally (though not consistently) has in view a contrast with analytic reason, not with rationality or reasonableness tout court.31 The interpretive advances in the book—the emphasis on Hume’s causal theory of assurance and on normative discriminations within custom (§2)— find themselves in the company of Kemp Smith’s mature idealism. Kemp Smith writes: “Hume speaks as if the ordinary conception of knowledge will remain unaffected, and will find in human nature a more secure foundation,” and does so “instead of formulating his thesis in [a] straightforward fashion, as the thesis that [in the moral and natural sciences] knowledge is not strictly knowledge at all, but only a humanly conditioned type of belief” (1941, 63). Why would this be the “straightforward” way for Hume to state his thesis? The answer is that Kemp Smith’s idealism, which blossomed only after the articles, takes Hume’s naturalism to imply that there is no such thing as empirical evidence (merely causes), no such thing as synthetic knowledge (merely opinion), and no genuine faculty of reason (merely instinct). This exhausts the “positive teaching” of Book I, from Kemp Smith’s idealist perspective. The book contains statements of Hume’s naturalistic reconstruction of reason (along the lines of the articles), statements to the effect that Hume debunks reason, and passages suggesting that these antithetical perspectives are internal to Hume’s own thinking.32 The antithesis is less between two veins in Hume’s texts than between Hume’s epistemological naturalism and Kemp Smith’s idealism. The latter masks and even relinquishes Kemp Smith’s recognition of Hume’s naturalistic reconstruction of reason. No surprise that commentators have paid little heed to this feature of the interpretation, focusing instead on the theme that belief is nonrational. Understanding Kemp Smith’s interpretive accomplishment requires that we factor out the corrosive influence of his idealism.

31. The interpretation of Hume in Mind must have developed explosively. It appeared just three years after Studies in the Cartesian Philosophy. That book’s chapter on Hume advances a fairly conventional interpretation (as Kemp Smith himself notes in 1905, 170n.1). For example, Hume is there portrayed as identifying the mind with sets of successive perceptions rather than a substance (cf. 1902, 231, 246–47), whereas in Mind the belief in the self is a natural belief. 32. Kemp Smith writes of “the stresses and strains to which [Hume’s] teaching is subject” (1941, 446). In the preceding paragraph, we find the naturalistic reconstruction of reason (albeit with an assist from reflection): “Even in the field of matters of fact and existence, [Hume] has allowed that reason in its reflective capacity is a useful, and indeed indispensable, ally; it assists the imagination to discharge its functions in a truly enlightened and reliable manner. There is, he maintains, a logic of such reasoning” (1941, 446; cf. 449). We also find, beginning with the next sentence, the alleged rejection of reason in the synthetic sphere: “Still, there is no denying the close kinship between Hume’s teaching and that of the skeptics. Has he not ousted reason from all supremacy in the domain of matters of fact and existence? Has he not treated causal

the kemp smith interpretation of hume

259

5. Kemp Smith’s Central Analogy Once the naturalistic reconstruction of reason assumes its rightful place in an inventory of theses within the Kemp Smith interpretation, an earlier question (§2) takes on new urgency. How could Hume claim to be giving a reconstruction of synthetic reason and empirical evidence if belief, like morality, is solely a matter of feeling or sentiment, and hence nonrational? We have seen that Kemp Smith himself recognizes this conflict within his interpretation (§§2, 4). My position is that Kemp Smith is mistaken in attributing to Hume the thesis that all belief is a matter of sentiment. A recurrent theme in Kemp Smith is that there is an “analogy” (1941, 391; cf. 119, 388), “a complete analogy” (1905, 339), between “belief” or “reason,” on the one hand, and taste, aesthetic and moral judgment, on the other. Belief and morality are “exactly on a level” (1941, 147; 1905, 343), “on all fours” (1941, 65). We shall see that the intended analogy is clear: Kemp Smith takes both the fundamental beliefs in the theoretical realm and aesthetic and moral judgments to result from an instinctive propensity to project internal feelings or impressions onto objects outside the mind. This thesis is in place in the articles, where the fundamental beliefs are center stage. Though Kemp Smith does not use the language of “projection,” he insists on the doctrine.33 He quotes the second Enquiry, Appendix 1 (1905, 342): “taste . . . has a productive faculty, . . . gilding or staining all natural objects with the colours, borrowed from internal sentiment” (EPM 294). In the case of necessary connection, he writes, citing Treatise I.iii.14 (T 167), of the “instinct which leads us to spread ourselves on external objects and to ascribe to them any internal impressions which they occasion in us” (1905, 162). In Kemp Smith’s view, projection also gives rise to the belief in “permanence and identity” (1905, 159). In the case of body, observing “gradual changes . . . leave[s] a feeling of sameness or identity”; “owing to the mind’s instinctive tendency to spread itself over external objects, and to ascribe to them any feeling they occasion,” we believe in “an objective identity in the real objects” (1905, 160).34 He interprets Hume on belief in the identity of the self along similar lines (cf. 1905, 161). Kemp Smith’s emphasis upon projection carries over to the book in all these cases (1941, 146–47, 198, for moral judgment; 119–20, 395–98, for necessary connection; 118, 476–78, for body; 556, for the self). In addition, the book goes beyond the articles in calling attention to Hume’s analogy in the Treatise (T 469) between virtue and vice and secondary qualities (1941, 198; cf. 28 and 1935, 32).

inference as a mode not of knowledge, but only of belief, operating through the imagination, with the uniformities of custom as its ultimate sanction?” (1941, 446). 33. For a history of the terminology within the Hume literature, see Kail 2001, 48n.1. 34. Kemp Smith takes himself to be discussing the belief in material substrata. The treatment in Mind, however, is applied in the book to the belief in body.

260

reflection and the stability of belief

Kemp Smith is wrong to think that the belief in permanence results from projection (§6). What matters here is that it is his thesis that the fundamental beliefs result from instinctive projections of internal feelings that secures the analogy between moral judgments and the fundamental beliefs in body and necessity (1941, 65, 147; cf. 1905, 339, 341–43) and thus animates his interpretive claim that belief is “subordinate” to passion. This helps explain why both in the articles and in the book Kemp Smith treats Hume’s general theory of belief—as a feeling that results from enlivening—as incidental to Hume’s position (§1). That belief is a matter of sentiment is a distinctive feature of the fundamental, projective beliefs. For Kemp Smith, the specific beliefs, which are custom-bred (§2), do not involve the projection of internal feelings. Kemp Smith does tend to generalize his thesis about the fundamental beliefs to a thesis about all belief or judgment (cf. 1941, 396).35 One contributing factor is that there is relatively little attention in the articles to the derivative beliefs (§2), which are due to custom, not projection. Another is that two passages, in I.iii.8 and I.iv.1, give some encouragement to the idea that all belief is solely a matter of taste or sentiment in the fashion of aesthetic and moral judgment. I believe these passages offer only superficial support for attributing to Hume a general sentimentalist theory of belief. Hume writes: “[A]ll probable reasoning is nothing but a species of sensation. ’Tis not solely in poetry and music, we must follow our taste and sentiment” (T 103). In context, his point is simply that probable reasoning results from custom, that it is derived “immediately” (103) from past repetition without any intervening argument—without “a moment’s delay,” “before we have time for reflection” (104; cf. 93).36 Similarly, when Hume writes that “belief is more properly an act of the sensitive, than of the cogitative part of our natures,” his point is again that it is “deriv’d from nothing but custom” (183), which is not “cogitative” in that it does not involve intermediating reflection. Treatise I.iii.8 and I.iv.1 are thus amenable to much more restrained readings.37 Further, attributing to Hume the thesis that all belief is a matter of sentiment generates a pressing objection to the course of Hume’s argument across Books I and III—as James Ward Smith demonstrated in 35. Walsh takes the idea further. He thinks that Hume’s view “really amounts to an attempt to explicate truth in terms of feeling. True belief is just belief in which the feeling of conviction occurs appropriately.” Walsh’s development of the idea (1970–71, 104–6) is subtle. 36. Dalrymple is correct to view Treatise 103 as an instance of “extravagant statements from Hume” (1986, 89). 37. Hume does appeal to the claim that belief is a feeling, rather than a “new idea,” to explain why it is not “in a man’s power to believe what he pleas’d,” why belief “depends not on the will” (T 624; cf. EHU 47–48). Kemp Smith holds that “since belief is precisely not subject to the individual’s arbitrary choice,” the imagination, in the sense of the general faculty of belief, “can be operative only when principles ‘permanent, irresistible, and universal’ are in control” (1941, 460). He thus might take Hume’s claim that belief is a feeling or sentiment to explain permanence and irresistibility, as well as involuntariness, in belief. One’s belief in a proposition, however, might respond to evidence, and hence not be permanent and irresistible, even though one cannot believe the proposition at will. Involuntariness does not require inevitability or unavoidability; we should not foist on Hume the view that it does.

the kemp smith interpretation of hume

261

1960.38 Hume claims that moral distinctions are derived from sentiment rather than reason; that morality is neither derived from a priori reason (T 463–68) nor “consists . . . in any matter of fact, . . . whose existence we can infer by reason” (468). Were all belief a matter of sentiment, then moral distinctions would be derived from sentiment even if they did arise from reasoning about matters of fact! Hume’s thesis that moral distinctions derive from sentiment rather than reason would lose its bite.39 Of course, one can insist that, for Hume, morality is a matter of sentiment, meaning that it is founded in impressions, and that belief is a matter of sentiment, meaning that it is due to instinct. The upshot of employing ‘sentiment’ this broadly is to obscure Hume’s project of providing a reconstruction of reason itself within instinct. Interestingly, Kemp Smith thought Hume made the mistake of assigning multiple uses to “feeling” and “sentiment.” In the book, when Kemp Smith proposes that for Hume all belief is a matter of sentiment, he raises a reservation: ‘[F]eeling’, . . . Hume’s general title for all the various manifestations of our ‘sensitive nature’, has, as he employs it, come to be an omnibus term used to cover a quite miscellaneous variety of mental experiences— the animal instincts, the passions both ‘direct’ and ‘indirect’, the moral and aesthetic sentiments, and, in the context of . . . Book I, also belief. (1941, 547; cf. 11–12, 44, 447) Such remarks prepare the ground for an “obvious criticism” (1941, 547): “‘Feeling’, ‘passion’, ‘sentiment’ are unanalyzed terms; and in dealing with 38. J. W. Smith 1960, 65, 69, and §III. 39. Barry Stroud tries to explain away the difficulty. Noting that in Book III causal reasoning falls on the “reason” side of the “reason-sentiment” contrast (cf. 1941, 162), he offers the following suggestion: “This is perhaps further evidence for Kemp Smith’s conjecture that Hume’s views on action and morals were composed first, before generalizing the point to all cognitive activity in Book I of the Treatise. The ‘reasoning’ that Hume here contrasts with ‘feeling’ or ‘passion’ includes ordinary causal reasoning, the conclusions of which he here regards as ‘calm and indolent judgments of the understanding’ (p. 457), whereas in Book I he holds the more radical view that even those conclusions do not belong to ‘reason’ or ‘the understanding’, but to ‘the imagination’” (1977, 263n.10). Even if Book III was drafted first, however, it was published after Book I. (I owe this point against Stroud’s suggestion to Stroud, in conversation.) J. W. Smith anticipates the reply that Books II and III were written before Book I (1960, 70–71). Also—as Kemp Smith recognizes (1941, 145n.2)—in the second Enquiry Hume retains the claim that morality is due to sentiment rather than to reason, including reasoning about matters of fact (cf. EPM 287–88). Stroud himself makes substantially the same point about the Enquiry (1977, 264n.10). In claiming that in (the later) Book I Hume removes causal reasoning from the understanding, Stroud overlooks Hume’s recurrent use of ‘reason’ in connection with experimental reasoning in Book I, his distinction between the imagination in a wide and narrow sense at Treatise 117n. (cf. 371n. in II.ii.7), where Hume equates reason with the former exclusive of the latter, and the entire project of providing a naturalistic reconstruction of normative distinctions at Treatise 225–26 and elsewhere (§2). [For a textual note on Treatise 371n., see “References to Hume,” this volume.] For Kemp Smith’s contention that main theses of Books II and III of the Treatise were conceived and composed prior to Book I, cf. 1941, 161, 173, 231, 245, 538.

262

reflection and the stability of belief

‘belief’, . . . [Hume] allows himself every advantage afforded by the laxities of this undefined usage. At need, belief can be spoken of as if it were merely feeling” (1941, 548–49). In sum, Hume’s “indiscriminate” (1941, 548) usage of ‘feeling’ and ‘passion’ evacuates his thesis about belief. Charity might have suggested that Kemp Smith reconsider.40 Kemp Smith attributes to Hume a projectivist account of the beliefs in body and necessity; on his interpretation, these beliefs are “accounted for in terms of a psychological mechanism in which impressions of reflexion have a chief role” (1941, 215; cf. 556). Hume is notoriously careless with terminology. When he writes of belief as a “feeling” (T 624; cf. 627) in the Appendix, he is not stating a doctrine, but avowedly groping for a theory: “This variety of terms [force, vivacity, solidity, firmness, and steadiness], . . . is intended only to express that act of the mind, which renders realities more present to us than fictions . . . Provided we agree about the thing, ’tis needless to dispute about the terms” (629/1.3.7.7). Yet, he provides four distinct arguments (625–27) against the view that belief is “some impression or feeling, distinguishable from the conception” or idea (625). In the course of the discussion, Hume writes that belief and mere imagination “are different to the feeling; but there is no distinct or separate impression attending them” (625). Belief is not an impression accompanying an idea, and a fortiori not an impression of reflection.41 Belief is a “feeling” in a broad or amorphous sense that does not support an analogy with the role of feelings, impressions of reflection, in the natural beliefs. I have called attention to an internal problem for Kemp Smith’s work on Hume: his claim that “All judgments of belief . . . are in [Hume’s] view based, like all judgments of value, exclusively on feeling” (1941, 397), insofar as it suggests that all belief is nonrational, is utterly antithetical to the naturalistic reconstruction of reason and of normative discriminations that he rightly identifies. The textual evidence for the subordination of all belief to passion is thin and readily neutralized. Attributing to Hume the view that all belief is a matter of sentiment works at crosspurposes to his claim that morality, unlike belief in matters of fact, is “founded” (T 546; cf. 574, 608) in sentiment. There is no concrete analogy between the account of belief and the role of projections of impressions of reflection in the moral and theoretical realms. Best to abandon Kemp Smith’s subordination thesis itself. The recognition of Hume’s naturalistic reconstructions of reason and of normative discriminations constitutes the core of what is worth preserving in the Kemp Smith interpretation. Attributing the subordination thesis to Hume is an interpretive error, another distraction—beyond Kemp Smith’s projecting his idealism onto Hume (§4)—from Hume’s naturalistic epistemology. The idealism contributes to this 40. J. W. Smith has a nice discussion (1960, 65 and §§II, IV). Also see Glathe 1950, 14; and Dalrymple 1982, 78–80, and 1986, 86–88. 41. Dalrymple cedes too much to Kemp Smith on this point (1986, 87).

the kemp smith interpretation of hume

263

mistake as well. In the inaugural lecture, all human values—spiritual and intellectual—“stand on the same plane of objectivity, and therefore justify parity of treatment” (1919, 15).42 Granted, for Kemp Smith values “have a more than merely human significance” (1919, 15), so that he could not abide a reconstruction of reason as due to the fabric and constitution of the human species, to instinct. At the same time, Kemp Smith welcomed the thought that Hume placed reason and morality “exactly on a level,” on the same plane, albeit the wrong one. This is a Hume who gets things half right. A highly general subordination thesis, that all judgment is a matter of feeling, thus exerted a strong gravitational pull on Kemp Smith. In this way, Kemp Smith’s idealism smoothes the path for his reading of Treatise 103 and 183. In the articles, Kemp Smith does not cite these passages; the sentimentalist theory of belief in the book is an accretion to the original interpretation (1941, 66, 387, 546, 551).43 Here we have another instance of Kemp Smith’s developed idealism spoiling his interpretation of Hume.

6. An Analogy Overlooked Much as Kemp Smith vastly overstates any tendency in Hume to assimilate all belief to feeling or sentiment, he ignores important similarities between synthetic reason and moral judgment within Hume’s philosophy. The source of this problem lies in Kemp Smith’s interpretation of Hume’s account of morality itself. For Hume, moral judgment depends on distinctive sentiments. Kemp Smith thinks these sentiments must be disinterested (1941, 167) and that Hume can allow their existence because “there are many disinterested propensities in our complex make-up” (1941, 140; cf. 27, 142–43, and 1905, 336). Taking this in conjunction with the claim that moral judgments are based “exclusively” or “solely” on feeling (1941, 13, 29, 197, 396; cf. 550), Kemp Smith has it that “moral approval or blame arises in the mind, . . . as a feeling to which we are immediately determined” (1941, 197). This is far from the case. It is sympathy that “produces our sentiments of morals” (T 577). Special resemblances and relations intensify feelings produced by sympathy (318), resulting in intrapersonal and interpersonal variation in sympathetic response. This leads Hume, in III.iii.1 (580–84, 590–91) and III.iii.3 (602–3), to consider an objection: how can sympathy, which is variable in its effects, be the foundation of moral judgment, which is invariable? Hume’s response is that “in order . . . to prevent those continual contradictions” resulting from variation in sympathy, we take up a “steady and general” point of view (581–82), a “general inalterable standard” (603; cf. EPM 229). We thus “correct” (T 582, 583, 585; cf. 603 and EPM 228) sympathy. Judgments reflecting sentiments 42. Kemp Smith writes in a 1919 letter: “[Samuel Alexander] wrote me, in acknowledging the Inaugural, that he regards my argument from absoluteness in knowledge to absoluteness in other spheres as the fundamental fallacy of idealism” (1981, 51). 43. Kemp Smith does, in Mind, quote Treatise 183 (1905, 157, 165).

264

reflection and the stability of belief

that arise within the steady and general point of view “are alone admitted in speculation as the standard of virtue and morality” (T 591). It is a matter of controversy how Hume would account for our access to the sentiments of approbation and blame within the steady and general point of view. Perhaps we employ imaginative projection or simulation. Perhaps we draw inferences about the sentiments that would arise were we to occupy the standard point of view. Either way, we rely on psychological or inferential mechanisms that supplement the operation of sympathy. Moral judgments do not derive from feelings that are a direct, immediate product of a disinterested instinct. Neither in the articles nor in the book does Kemp Smith discuss Hume’s doctrine of corrections to sympathy, or even Hume’s raising variability in sentiment as an objection to sentimentalism.44 These are stunning and ironic omissions, given Kemp Smith’s contention that “it was through the gateway of morals that Hume entered into his philosophy” (1941, vi; cf. 110, 538). How are they to be explained? T. H. Green and Edna Aston Shearer had discussed Hume on corrections in work Kemp Smith read.45 With respect to synthetic belief, belief in matters of fact, Kemp Smith is of two minds— alternately taking Hume to reconstruct reason and to discredit it (§4). He is not similarly conflicted with respect to Hume’s account of morality: “[T]here is, on Hume’s theory . . . , no such thing as moral obligation, in the strict sense of the term. There is, that is to say, no intrinsically self-justifying good that with authority can claim approval. The ultimate verdict rests with the de facto constitution” of the individual (1941, 201; cf. 27–28, 43–44) or human species (1941, 201–2). We again encounter Kemp Smith’s idealist commentary on Hume, at this point as offering a debunking account of morality. Yet, this cannot be the entire explanation for Kemp Smith’s failure to discuss corrections to sentiment. Kemp Smith considers details of Hume’s account of synthetic reason—for example, the distinction between good and bad customs—even though he thought a naturalistic theory of reason hopeless (§§2, 4). In part, Kemp Smith was simply much less interested in ethics than in the understanding.46 He chose both to translate and write a commentary on Kant’s Critique of Pure Reason, not Practical Reason. His own work outside the history of philosophy pursued metaphysics and epistemology. More deeply, the doctrine of correction calls into question Kemp Smith’s characterization of natural belief. In his discussions of corrections, Hume 44. In Mind, there is only one reference to III.iii (1905, 346)! In the book, there are fewer than ten citations to III.iii.1 and III.iii.3 (1941, 151, 154, 167, 174, 175). None of these addresses variability or corrections. Kemp Smith does introduce a notion of “correction” in his discussion of I.iv.2 (1941, 454). Norton observes that Kemp Smith is also silent on the role of corrections in Hutcheson (1982, 68–69, 75, 128–30). 45. See Green 1886, §§52–54; Shearer 1915, 74–83; and Kemp Smith’s bibliography at 1941, ix. 46. The second of the articles in Mind, examining Hume’s naturalism in ethics, is half the length of the first. Laird wrote in a review of Kemp Smith: “Substantially, however, and inevitably, it is Hume’s epistemology and not his ethics that is the theme of this book. Over fivesixths of the 566 pages are devoted to that topic” (1941, 129). Ewing observes that Kemp Smith’s “account of Hume’s ethics is scrappy in places” (1942, 75).

the kemp smith interpretation of hume

265

allows that the spontaneous products of sympathy can be “stubborn or inalterable” (T 582; cf. 583, 585, 603, and EPM 227–28). This is awkward for Kemp Smith’s emphasis on inevitability and irresistibility as the source of nature’s sanctions (§3). According to Hume, effort is required if we are to reach the normatively preferred judgments within the steady and general point of view; most problematically for Kemp Smith’s interpretation, spontaneous sentiments resist correction but lack normative standing.47 Kemp Smith’s achievement is the recognition that Hume proposes a naturalistic reconstruction of normative discriminations within synthetic reason (§2). Not any belief based on custom constitutes “experience in [the] normative sense.” The full normative standing of the products of custom requires that its more spontaneous operations be refined or processed, subjected to “control” (1941, 128). Kemp Smith’s failure to engage the doctrine of corrections to sympathy prevents him from generalizing this interpretive insight. Not any sentiment of approval or disapproval arising from sympathy is a moral sentiment. The full normative standing of the products of sympathy requires that its more spontaneous operations be controlled or corrected. The outputs of basic instinctive propensities—whether custom or sympathy—are subject to control and correction. The domains of synthetic belief and judgments of approval are parallel in that both admit of normative discriminations.48 It would be well if we could locate a Humean theory of acceptable or privileged judgment operative across Books I and III. Since moral judgment is not due to reason (§5), including experimental reasoning about matters of fact, causal inference cannot play this role. The theory that predominates in Kemp Smith, cast in terms of irresistibility and inevitability, runs into the problem that uncorrected products of sympathy can be irresistible. Other candidates include the naturalistic theories in terms of adaptiveness or proper functioning that descend from Kemp Smith (§3). Reflective endorsement theories (§3) are also a possibility, though a special case of a stability interpretation (in that beliefs that do not survive reflection are liable to be unstable). I will focus on this last option, which takes a cue from Hume’s response to the objection from variability, the material in Book III that Kemp Smith ignores. The apparent conflicts owing to variation in sympathy motivate us to adopt “a more stable judgment” (T 581) with reference to a “steady and general,” or standard, point of view. Stability also has a role in explaining 47. Here we find pressure within the Kemp Smith interpretation to soft-pedal the doctrine of corrections. Thus, Stroud writes that the doctrine “puts considerable strain on [Hume’s] official theory of the role of feelings in morality” (1977, 189). But though morality is a matter of sentiment, not any sentiment is a moral one. See Norton 1982, 151n.67, and my 2002, §IV.6. Stroud denies Hume the resources of corrections while lamenting the “relative crudity” (1977, 186) of Hume’s account of moral judgment. 48. There is in Kemp Smith a passing suggestion of this parallelism: “For Hume, . . . logic and ethics rest on one and the same basis: experience, as extended in and through our reflective activities, is normative for both” (1941, 388). As far as I can see, however, the only contribution of reflection or reason Kemp Smith recognizes (in the case of the natural virtues) is that of a servant, in the selection of suitable ends. See especially his chs. 6 and 9.

266

reflection and the stability of belief

why one particular point of view, rather than others, constitutes the standard. Hume writes: “we cannot afterwards fix ourselves so commodiously by any means as by a sympathy with those, who have any commerce with the person we consider” (583; cf. EPM 218).49 The point of view of the agent’s “narrow circle” (T 602; cf. 582, 590, 591) recommends itself—in virtue of its psychological salience—for a stabilizing role. Hume intends to explain why the point of view on which we converge is normatively privileged. The standard point of view is a distinctively moral perspective because it is uniquely able to stabilize judgments of approval and disapproval. This is a special case of an overarching idea—normative pride of place derives from mechanisms that stabilize judgment. This perspective illuminates the centrality of causal inference in theoretical judgment. Causal inference is privileged because custom or repetition infixes belief and tends to be stabilizing. Instabilities nevertheless arise within perception, memory, and causal inference. A stability interpretation thus also illuminates Kemp Smith’s chief examples of bad customs (§2). If mere recency of an experience influences belief, our degree of assurance will vary over time (T 143); if we allow all regularities equal standing, we will encounter conflicting regularities in the form of inductive inconsistencies (149–50). These are the second and fourth kinds of unphilosophical probability, respectively, in I.iii.13. Hume there looks to second-order principles, “general rules” (150; cf. 141, 632/1.3.10.12), for correcting, and thus stabilizing, beliefs based on memory and on custom. These are Kemp Smith’s “wider and more reliable forms of custom” (1941, 95). Similarly, Hume invokes stabilizing “general rules” in III.iii.1 (T 584–85), to deal with the problem of “[v]irtue in rags.” Although Kemp Smith introduces the problems of recency and conflicting customs, he does not cite I.iii.13 in this regard; his book neither quotes the section nor mentions “unphilosophical probability” by name.50 This is of a piece with Kemp Smith’s inattention to the parallel doctrine of corrections in Hume’s moral theory. Donald MacNabb was perhaps the first to stress the systematic role of stability in Hume—in his 1951 book, published ten years after Kemp Smith’s. Whereas Kemp Smith was preoccupied with belief as a feeling, MacNabb was alert to strands in Hume that characterize belief in terms of fixity or steadiness. MacNabb then appeals to destabilizing conflicts in order to elucidate Hume’s

49. See my 2003 [this volume, ch. 8] for discussion and refinements. 50. There are but two passing notes citing I.iii.13 (1941, 249n.4, 427n.3). Kemp Smith is aware of a doctrine of “general rules” in I.iii.13, but he only considers its application to a problem in I.iii.12 (1941, 427). Although he devotes a chapter to the probability of chances and of causes, Kemp Smith omits unphilosophical probability in his diagram of Hume’s division of human reasoning into knowledge and the various species of probability (1941, 415). (Stroud 1977 also makes no mention of I.iii.13. See also note 47.) The topic of general rules had received only passing notice prior to Hearn’s articles in 1970 and 1976. Indeed, discussions of unphilosophical probability had been fleeting, prior to MacNabb 1951/1966, 96–99, and Passmore 1952/1968, 59–64. See Hendel 1925, 203–4 (or 1963, 170–71), and Laird 1932, 91.

the kemp smith interpretation of hume

267

accounts of why some judgments of probability are rejected as “unphilosophical” and why some judgments of approval are not distinctively moral.51 There is an additional element in MacNabb’s interpretation: Hume associates instability with felt uneasiness, discomfort, dissonance—an unpleasant feeling.52 Indeed, evidence is scattered across all three books of the Treatise (cf. T 205–6, 215, 453, 592–93; also cf. Essays 60–61) that in virtue of a general aversion to pain, the uneasiness in instability or conflict provides a natural motive to adjust belief or judgment. Perhaps Hume’s thought is that we feel uneasy in the variation in judgments of approval that results from untutored sympathy; similarly, we feel uneasy in the variation in confidence or assurance that results from the unrefined products of memory and causal inference.53 On the stability interpretation, Hume’s position is that the normative standing of judgments in both the theoretical and moral realms is naturalized as deriving from the motivational force of the felt uneasiness in unstable doxastic states. Of course, these brief remarks are not intended to sustain MacNabb’s line of interpretation. Comparison to Kemp Smith’s account of I.iv.2, home territory for his interpretation, is instructive. Kemp Smith locates nature’s sanctions in the instinctual projections of internal feelings. In the case of the belief in body, the textual evidence that Hume appeals to the projection of a feeling of identity is negligible. Kemp Smith relies on a comment within Hume’s discussion of the constancy of perceptions (1941, 477; cf. 118): in considering a succession of invariable but interrupted objects, “The passage from one moment to another is scarce felt” (T 203).54 All Hume means by this is that there is “an easy transition,” a “smooth and uninterrupted progress” (204). Even if the effortless transition (cf. 254) involves a feeling, Kemp Smith offers no evidence that Hume appeals to a projection of the feeling to explain the belief in body. To the contrary, Kemp Smith notes that the belief arises from the mind’s propensity to “mistake” (1941, 476; cf. 481) the idea of an invariable and interrupted object with that of an object that is invariable and uninterrupted. To confuse one idea for a related idea is not to project a feeling of identity. By contrast, the inclination to attribute identity to interrupted impressions, coupled with the recognition that the interrupted impressions are not identical, leads to “sensible uneasiness” (T 205), so that the mind is “uneasy” (206). The vulgar response, attributing continued existence to perceptions, is “contrary to

51. For belief, see MacNabb 1951/1966, 69–81; for stability, 80, 94–100, 191–93, esp. 99n.1. There is a glimmering of the stability option in Kemp Smith 1941, 493–94, 543. I have developed MacNabb’s ideas in my 2002, esp. chs. 3–4. 52. MacNabb 1951/1966, 99, 126, 192. 53. I discuss both applications in my 2002, ch. 4, especially §§IV.4–5, and 2004, §§I.1 and IV.1. Korsgaard stresses this point in application to Hume’s moral theory (1999, 24–25). 54. Kemp Smith generalizes the idea to the belief in mental continuity or a continuous self (1905, 161; cf. 1941, 96–99, 499–505, 556), though he neglects a slightly better passage for his purposes. In discussing a highly analogous mechanism in the case of belief in the mind’s identity, Hume writes that consideration of an uninterrupted and invariable object and consideration of a succession of merely related objects “are almost the same to the feeling” (T 254).

268

reflection and the stability of belief

the plainest experience” (210). This “contradiction” (206, 215) and “opposition” (205, 215) lead to the philosophical belief in double existence, another attempt to “set ourselves at ease” (215, twice). This is not to say that the vulgar belief or philosophical system is justified; responses to instability are often themselves conflicted, so that it is sometimes preferable to root out an instability at its source than to attempt to resolve it.55 What matters is that a feeling of uneasiness, unlike a feeling of identity, plays a prominent role in I.iv.2. Though Kemp Smith is unsympathetic with instinctive beliefs (§4), he found instinctive fundamental beliefs (§2), the beliefs in continuants and causality, congenial insofar as he viewed them as precursors of Kant’s a priori concepts. “[T]he natural beliefs,” he writes, “create for us the world of ordinary consciousness” (1941, 224; cf. 124, 222). (One chapter of Kemp Smith’s Prolegomena is given over to the categories of substance and causality.) Kemp Smith strives to find a similarly parallel treatment of the fundamental beliefs in Hume. He thus seeks a feeling apt for projection, along the lines of a feeling of determination in the context of the belief in necessity (T 156, 165, 169; EHU 78n.), to ground the Humean account of body. The uneasiness passages, which are of no help in this regard, receive no play.56 Let us retrace a step or two. Kemp Smith ought to have generalized his core insight about Hume’s reconstruction of synthetic reason: Hume intends to provide naturalistic grounds for normative discriminations among judgments of approval as well as judgments of fact, in terms of a general theory of privileged judgments. Had Kemp Smith been more engaged with the accounts of the natural virtues and of unphilosophical probability, he might have recognized the importance of stabilizing corrections in Hume. An interpretation highlighting the motivational role of uneasiness in instability indeed has affinities with that of Kemp Smith, broadly construed. Much as moral obligation to perform an action requires a natural motive, obligation to accept a judgment rests in natural motives to avoid uneasiness. These feelings of uneasiness are not projected or spread onto the world. Nor is uneasiness a component of belief, of theoretical or practical judgments themselves. It is an aversive feeling that accompanies sets of conflicting judgments, motivates their revision, and is the source of normativity in judgment within Hume’s system. If so, the feeling of uneasiness in instability plays a crucial role across Books I and III. Interpreting Hume in light of stability and uneasiness vindicates Kemp Smith’s emphasis on impressions of reflection (§5), though not his insistence that Hume upends reason in favor of feeling.57

55. See my 2002, §§V.1–5. 56. One relevant passage is buried in a lengthy quotation at 1941, 490. 57. I read a version of this chapter at “Kemp Smith 101, A Colloquium: 101 Years of ‘The Naturalism of Hume,’” Oxford University, May 19–20, 2006. I am grateful to the other speakers— Peter Kail (also the conference organizer), Peter Millican, Barry Stroud, and John Wright— for helpful comments and discussion. David Velleman generously provided detailed comments leading to many improvements. Special thanks to Patricia Boyd and the Special Collections staff at the University of Edinburgh Library for their help with access to Kemp Smith’s papers.

the kemp smith interpretation of hume

269

10 Psychology, Epistemology, and Skepticism in Hume’s Argument about Induction

1. Introduction do we yet understand Hume’s project in his main argument about induction in Treatise I.iii.6? Is the argument skeptical? Is it even so much as epistemological, skeptical or otherwise? Or is it merely psychological? Although the literature has made much progress on these questions, we have not reached a full understanding of Hume’s position. I hope to fill in some missing pieces of the interpretive puzzle.

2. The Skeptical Interpretation of I.iii.6 According to the skeptical interpretation of I.iii.6, the section advances the problem of induction, in the fashion of Russell in the Problems of Philosophy.1 This interpretation, which held sway for decades in the middle of the twentieth century, has its attractions.2 It is philosophically interesting. Also, it coheres with the Beattie-Reid tradition of interpretation, on which Hume is utterly destructive, an arch-skeptic with respect to causal necessity, the causal principle (that every new existence, or new modification of existence, has a cause), the external world, the substantial self, and so forth.3 Finally, 1. Russell 1912, ch. 6. 2. For some prominent examples, see Prichard’s 1932 lectures on Hume; Russell 1945; Will 1947; Popkin 1951b; Ayer’s 1960 Oxford inaugural lecture; Flew 1961; Stove 1965; Bennett 1971; Popper 1972; Hacking 1975; and Stroud 1977. The Prichard lectures are in his 1950 and the Ayer lecture in his 1963. 3. This is the interpretation Kemp Smith attacked (1905 and 1941).

270

the skeptical interpretation, unsurprisingly, has some basis in the text. Section I.iii.6 does supply key premises of the skeptical problem: first, inductive inference presupposes that nature is uniform; second, there is no demonstrative argument to show that nature is uniform; and third, no probable argument could show that nature is uniform, without begging the question. Hume would also seem to draw the argument’s conclusion, that there is no justification whatsoever for belief in the uniformity principle and hence for inductive inference: he writes that there is “no reason” (T 92; cf. 91, 139) to draw an inference from the unobserved to the observed. In order to assess the skeptical interpretation, it is necessary to insulate the question of whether Hume is a skeptic specifically with respect to induction from his overall skeptical tendencies. No one denies that Hume at least takes us near the brink of deeply skeptical results.4 He argues that our faculties lead to contradictory beliefs about the existence of matter (T 231; EHU 154–55).5 More dramatically, Hume writes in the concluding section of Book I: “I am ready to reject all belief and reasoning, and can look upon no opinion even as more probable or likely than another” (T 268–69). This seems to imply rejection of beliefs about the unobserved as a special case, and on the ground that all beliefs are equally probable, presumably because they are not justified at all. At the same time, it is important to distinguish among different lines of argument to the latter conclusion. Incoherence in our faculties, for example, might be a ground for skepticism, but it is a ground distinct from the Russellian problem of induction.6 The issue before us is whether in I.iii.6 Hume advances skepticism on the basis of this problem. N. S. Arnold and Janet Broughton independently produce a consideration that should give pause.7 In the I.iv.7 buildup to his readiness to reject all belief, Hume relies on a series of what he takes to be prior results: the contradiction about matter in I.iv. 4 (T 265–66), the discovery in I.iii.14 that causation involves no “connexion” or “tie” outside the mind (266–67), and a “dangerous dilemma” arising from the claim in I.iv.1 that the understanding subverts itself (267–68). Whereas Hume’s own footnotes cite the three earlier sections, he neither provides a note to I.iii.6 nor relies on the argument of that section. If Hume arrives at skepticism on the basis of considerations advanced in I.iii.6, surely he would appeal to them in the concluding section of Book I. Not only does Hume not refer to I.iii.6 in the course of his skeptical tailspin in I.iv.7, Hume’s project in the Treatise, his argumentative practices, and his explicit statements about induction commit him to the thesis that inductive inference is justified. The evidence against the skeptical interpretation

4. Although see M. Williams 2004. 5. [In this chapter, quotations of the Treatise, Abstract, and first Enquiry are based on the volumes edited by Norton and Norton (NN) and Beauchamp (B-00) in the Clarendon edition of Hume’s works.] 6. Noonan is clear about this distinction (1999, 129–31). 7. Arnold 1983 and Broughton 1983.

hume’s argument about induction

271

has steadily accumulated since the 1970s, to the point that the interpretation has effectively been demolished.8 This is not to say there are no holdouts, but that holding out is now untenable.9 It would be well to flesh this out a bit. First, Hume’s project. The Treatise carries the subtitle “being an attempt to introduce the experimental method of reasoning into moral subjects” (T xi). The result, in part, is the elaboration of a variety of associative mechanisms: for example, association by the relation of cause and effect, in Book I; the double association of impressions and ideas, in Book II; and the associative mechanism of sympathy, in Books II and III. Hume pursues the science of human nature across all three books of the Treatise. Second, Hume’s argumentative practices in support of this project. In the initial section of the work, Hume relies on inductive evidence to establish his principle that every simple idea exactly resembles a preceding impression (T 4–5). Just two sections following the main argument about induction, Hume appeals to “experience” and “experiments” to confirm—indeed, “to prove”— his associationist account of belief (99, 100). Many subsequent sections are replete with experiments in support of Hume’s psychological hypotheses. Book II includes a section “Experiments to confirm this system,” that is, Hume’s system of the indirect passions. Third, Hume’s explicit statements about induction. Hume tends to assimilate inductive inference to causal inference and thus often uses the latter as a stand-in for the former.10 Hume writes of causal inference to belief in the unobserved as leading to epistemic success. In I.iii.2, causation is the only relation that enables the mind to “go beyond what is immediately present to the senses, either to discover the real existence or the relations of objects” (T 73, emphasis added); the relation of causation “informs us of existences and objects, which we do not see or feel” (74, emphasis added). There are similar passages subsequent to I.iii.6. In I.iii.8, a person who stops his journey at a river “foresees the consequences of his proceeding forward; and his knowledge of these consequences is convey’d to him by past experience” (103, emphasis added; cf. 104, 148); in I.iii.9, the relation of cause and effect “brings us acquainted with such existences, as . . . lie beyond the reach of the senses and memory” (108, emphasis added). In the Abstract, “No matter of fact can be proved but from its cause or its effect. Nothing can be known to be the cause of another but by experience” (654, emphasis added).11 Similarly, Hume writes: “cause and effect . . . ’tis the only [connection or relation of 8. For some early inventories of the evidence, see Beauchamp and Mappes 1975; Connon’s 1976 conference paper on Hume’s naturalism; Winters 1979; and Beauchamp and Rosenberg 1981, as well as Arnold 1983 and Broughton 1983. Connon was published in 1979. 9. For attempts to resist the case against the skeptical interpretation, see Penelhum 1992 and Winkler 1999. 10. Price 1940, 29, and 1969, 176–79; Passmore 1952/1968, 29–34; and Pears 1990, 71–72. 11. Much as causal inference constitutes epistemic success in the Treatise, we have in the Enquiry: “Had not the presence of an object instantly excited the idea of those objects,

272

reflection and the stability of belief

objects], on which we can found a just inference from one object to another” (89, emphasis added). This endorsement occurs within Hume’s main argument about induction. At I.iii.13, causal inference is “just and conclusive” (144). The striking continuity in Hume’s positive evaluations of inductive inference throughout Part iii is difficult to explain on the hypothesis that I.iii.6 constitutes a skeptical turning point. What is more, even after I.iii.6 Hume recognizes gradations in inductive evidence. At I.iii.13, “Of unphilosophical probability,” he provides an inventory of “degree[s] of evidence” (T 153–54) that includes proofs and also probability, beliefs based on observation of conjunctions that are not constant or on infrequent observation of conjunctions (130–31, 142). Within probability, there are degrees of “force” (130) and “evidence” (131, 154). Hume devotes I.iii.13 to “unphilosophical probability” (143), in contrast to “kinds of probability [that] are receiv’d by philosophers” (143).12 All this, even though the problem of induction levels all inductive inference. There is extensive evidence that Hume endorses inductive inference.13 How might proponents of the skeptical interpretation of I.iii.6 respond? Russell thought Hume simply inconsistent.14 Barry Stroud attempts to transmute this inconsistency into the very message of Hume’s philosophy. In Stroud’s hands, Hume is purposefully calling attention to the inevitability of the inconsistency, for himself and everyone else, given the universality and irresistibility of inductive inference. Stroud writes: “Hume’s theory taken all together shows that most of our beliefs must be wrong or unreasonable; given the way we are, we could not have those beliefs unless that were true.”15 Hume is making a point about an epistemic feature of the human condition. Stroud’s interpretation is much more interesting than Russell’s charge of inconsistency, but it equally presupposes that Hume is a skeptic about induction. We have seen that this cannot be correct.

3. The Descriptivist Interpretation My critical focus will be on a different and more influential line of interpretation, which I dub descriptivism.16 The descriptivist reading grants that Hume

commonly conjoined with it, all our knowledge must have been limited to the narrow sphere of our memory and senses” (EHU 55, emphasis added); and this: “The existence . . . of any being can only be proved by arguments from its cause or its effect; and these arguments are founded entirely on experience” (EHU 164, emphasis added). 12. Similarly, in the Enquiry, some events are “more probable” (EHU 58) than others; and, in a passage emphasized by Meeker (1998, 38), “A wise man . . . proportions his belief to the evidence” (EHU 110; cf. 117). 13. My 2008, 112–14, provides detail and, especially at item (g), additional kinds of evidence. 14. Russell 1945, 672. See also Flew 1986, 56–57. 15. Stroud 1991, 276. 16. Noonan calls this view “literalism” (1999, 119–28).

hume’s argument about induction

273

does not advance skepticism about induction in I.iii.6, but takes this to be a symptom of the absence of normative epistemology in the section. Hume does not there engage in either negative or positive epistemological evaluation of induction. Rather, I.iii.6 is wholly given over to the causes of inductive inference; it is an exercise in cognitive psychology. Of course, Hume’s investigation may result in fairly high-level or “theoretical” conclusions. The ‘descriptivist’ label does not signal a contrast between low-level observation and higherlevel theory, but rather one between empirical and normative inquiry. Any viable version of descriptivism will have to concede that Hume in some sense presupposes the justification of inductive inference. After all, he relies on inductive inference throughout Part iii. According to the descriptivist, Hume’s psychological investigation does not undermine the commitment to the legitimacy of inductive inference. This is because I.iii.6 does not so much as address the epistemic status of inductive inference. The psychological study of that section is neutral with respect to normative questions. In order to get descriptivism off the ground, its proponents must neutralize the passages in which Hume says there is “no reason” for inductive inference to the unobserved. One common descriptivist strategy proceeds in two steps. The first is to construe the “no reason” passages as somewhat awkward formulations of the claim that inductive inference is not caused or produced by reason. The second step is to take this claim to be directed ad hominem at a conception of “reason” Hume himself rejects. Descriptivists can disagree on whether Hume’s point is directed at Cartesian, or Lockean, or perhaps any nonassociative conception of reason.17 The important point is that the claim that inductive inference is not due to “reason” in a specialized sense need not commit Hume to the claim that inductive inference is unjustified; the negative psychological claim about the causation of inductive inference leaves open the question of justification. A compelling and relatively pure version of the descriptivist interpretation may be found in the paper by Broughton mentioned earlier. More recent literature generalizes the interpretation beyond I.iii.6. The idea is that the whole of Part iii of Book I is purely descriptive. It is not until Part iv, perhaps quite late in Part iv, that Hume undertakes any epistemological evaluation of inductive inference. On the most promising version of this position, due to Don Garrett, Part iii is descriptive of the psychology of epistemic evaluation and not itself normative.18 This generalization of a descriptivist reading beyond I.iii.6 has an important advantage. It provides a foothold for explaining away the implicit and explicit commitments to inductive inference, both before and after the main argument about induction; they are merely part of the data for psychological study. Much as psychology can investigate inductive inference, it can investigate why we approve of inductive inference 17. For some variants, see Garrett 1997, ch. 4; Millican 2002, §§1, 10.3; and my 2008, 110–11. Although he is a descriptivist, Garrett does not himself accept this approach. 18. Garrett 1997, esp. chs. 4, 7, 10. For Owen, I.iii is descriptive of epistemic distinctions internal to a practice (1999, chs. 6, 8, 9).

274

reflection and the stability of belief

(where we do approve of it). At least in Part iii, however, these commitments do not issue from Hume qua epistemologist—so the descriptivist tells us. Even a generalized descriptivist interpretation faces serious difficulties. Aware that descriptivism constitutes a large-scale interpretation with considerable resources to ward off criticism, I sketch a few of the problems. In the first place, whether or not Hume operates with a specialized conception of “reason” within I.iii.6, Hume applies ‘reason’ and its cognates to causal inference in I.iii.6 subsequent to the main argument about induction (T 93), in I.iii.7 (94, 95, 97n.), and also in I.iii.8, after attributing causal inference to custom (103, 104, 105). This terminology would seem to carry a positive epistemic connotation. Descriptivists can contend that, following I.iii.6, ‘reason’ simply refers to whatever faculty causes inductive inference.19 Let me offer one consideration that tells against this. In I.iii.9, Hume distinguishes within the faculty of association between “reason,” which includes “all probable reasonings,” and the “imagination,” which includes “whimsies and prejudices, which are rejected under the opprobrious character of being the offspring of the imagination” (117–18n.).20 Earlier in the section, within his discussion of two systems of realities—one based on perception and memory, the other on causal inference— Hume contrasts beliefs “arising from custom and the relation of cause and effect” with beliefs that “are merely the offspring of the imagination” (108). Similarly, in I.iii.13 the kinds of probability received by philosophers are “reasonable foundations of belief and opinion” (143, emphasis added). Further, the I.iii.9 distinctions anticipate that in I.iv.4, between the “permanent, irresistible, and universal” principles—“such as the customary transition from causes to effects, and from effects to causes”—which Hume there accepts, and the “principles, which are changeable, weak, and irregular,” which “are observ’d only to take place in weak minds” (225). I think it difficult to resist reading the occurrences of ‘reason’ and its cognates after I.iii.6 as honorifics.21 In the second place, the descriptivist interpretation must strain to accommodate a number of features of Hume’s positive evaluations of inductive inference. There is the sheer mass and coherence of these positive assessments—inductive inference as constituting an epistemic success, as just, and as generating degrees of evidence.22 Garrett suggests that, were Hume not to offer such evaluations, “he would provide a counterexample to his own psychological theory,” which “arguably . . . entails that we will, on the whole, continue to approve epistemically of our engaging in [such inference], so long as it continues to succeed.”23 This is not 19. Cf. Garrett 1997, ch. 4, 1998, 178–80, and 2002, Appendix. 20. The term ‘probability’ has a narrower meaning at Treatise 124. 21. See my 2002, §II.3, and Millican 2002, §12. 22. For skirmishes with descriptivist treatments of some of the texts, see my 2002, 43–44nn.12–13, 47n.19, 102nn.2–3, 131n.43. 23. Garrett 1997, 79–80; cf. 157–58. Similarly, Fogelin appeals to Hume’s thesis that we are psychologically compelled to undertake epistemic assessments in order to explain why the positive evaluations of inductive inference do not undermine the skeptical interpretation of I.iii.6 (1985, 148–49).

hume’s argument about induction

275

persuasive. Even if approval of causal inference is unavoidable, putting such commitments in writing is not. Further, some of Hume’s evaluations are second-order. Hume writes at I.iii.13: “We shall afterwards[*] take notice of some general rules, by which we ought to regulate our judgment concerning causes and effects” (T 149). Hume’s footnote references I.iii.15, “Rules by which to judge of causes and effects,” where he offers eight rules, “all the logic I think proper to employ in my reasoning” (175). It is a stretch to suggest that Hume is merely reporting that we are psychologically determined to engage in and approve of higher-order evaluations.24 In addition, Hume does not say things he ought to be expected to say on the descriptivist interpretation. Hume does not pause to note that his own positive evaluations are offered as nothing but part of the data for psychological study. Similarly, when Hume utilizes inductive inference—appealing to “experiments” and the results they “prove”—he does so unflinchingly, without cautionary comment, in each Book of the Treatise. In the third place, we have observed that Hume writes that inductive inference is “just” not only in I.iii.6 itself (T 89) but also later in I.iii.13, “Of unphilosophical probability” (144). Beyond that, Hume applies ‘just’ and its cognates to inductive inference well into Part iv: in I.iv.2, “Of skepticism with regard to the senses” (216) and in I.iv.4, “Of the modern philosophy” (225). The latter passage, where Hume is giving an example of the operation of the permanent, irresistible, and universal principles, is well-known: “One who concludes somebody to be near him, when he hears an articulate voice in the dark, reasons justly and naturally; tho’ that conclusion be derived from nothing but custom” (225).25 What is the descriptivist reading to make of the continuity in Hume’s seemingly positive assessments of causal inference across Parts iii and iv of Book I?26 These assessments occur as late as I.iv.4. If Hume is not yet engaged in normative epistemology there, when does this enterprise set in? Sections I.iv.5 and I.iv.6, “Of the immateriality of the soul” and “Of personal identity,” are by and large given over to metaphysics. Descriptivism seems to require that cognitive psychology does not give way to normative epistemology until the final section of Book I.27 This seems an odd role for a section entitled “Conclusion of this book.” The alternative is to concede

24. For these points, see my 2002, 45n.15. 25. There is also continuity in Hume’s positive assessment of causal inference across the Treatise and the Enquiry. In Sections 10 and 11, “Of Miracles” and “Of a Particular Providence and of a Future State,” there are a half dozen references to “just” reasoning or reasoners, or to reasoning “justly” (110, 113, 136, 139, 142, 145)—these in a context where Hume is relying on his distinction between legitimate and illegitimate forms of causal inference to criticize arguments for God based on miracles and design. See also notes 11 and 12. 26. For this question, and its bearing on the descriptivist contention that Hume’s use of ‘just’ and its cognates in Part iii is not normative at all, see my 2002, 43n.12, 43–44n.13. 27. Hence Garrett 1997, 95, 230, 232. Cf. Owen 1999, esp. 205–6.

276

reflection and the stability of belief

that the Treatise contains normative epistemology “all the way back,” that is, at least as early as Part iii.28

4. The Epistemic Character of I.iii.6 Let us put to the side the difficulties I have canvassed for the generalized descriptivist interpretation. Descriptivism cannot survive considerations internal to I.iii.6 itself. All descriptivists are committed to the thesis that Hume is not engaged in normative epistemology in that section. According to the interpretation, Hume’s point is merely psychological, that reason does not cause or determine us to make inductive inferences; rather, inductive inferences are caused by some other faculty. This is a negative conclusion within cognitive psychology. On the descriptivist interpretation, I.iii.6 has no epistemic force at all. Against this, I will argue that Hume does make a negative epistemic point, albeit one that falls short of radical skepticism. I begin by noting an ambiguity that tends to beset the descriptivist reading. Consider a formulation of Garrett’s: “[Hume] is arguing that we do not adopt induction on the basis of recognizing an argument for its reliability.”29 Are advocates of the descriptivist reading supposing that Hume restricts the arguments he is canvassing to good arguments, or must he also be on the lookout for bad arguments that seem to be good ones and thus manage to cause inductive inference?30 The ambiguity here is Hume’s own. In both I.iii.6 (T 89) and the corresponding section of the first Enquiry (EHU 32, 37, 38), he uses the terminology of “founded” and “foundation” without explicitly saying whether he is looking for a psychological or an epistemic foundation.31 Hume does not pause—not even at critical junctures—to distinguish merely psychological and epistemic notions. The interpretive issue depends on two cases. First, is Hume’s point that no argument that purports to be demonstrative causes inductive inference or 28. Why would Hume defer normative epistemology until after Part iii? Garrett sometimes portrays Hume as engaged in a Quinean enterprise, with the cognitive faculties investigating themselves (1998, 186). Garrett has suggested (in conversation) that Hume withholds his epistemological evaluation until all the empirical data are in. It would be surprising, however, if Hume thought that he could ever possess all the relevant psychological data. In any event, why would he not provide provisional epistemic assessments of inductive inference and other beliefforming mechanisms as the empirical investigation proceeds? Cf. my 2002, 14–15, 88–92, and 2004, §II.2. 29. Garrett 1997, 92. Garrett now puts no weight on an argument specifically for the reliability of inductive inference (2002, 333n.30). 30. The same question could be asked of formulations of my 2002, §II.2, and my 2008. 31. This ambiguity, perhaps even vacillation, is but one instance of Hume’s tendency to use terminology that straddles purely psychological and epistemic notions. For example, he writes that causation is the only relation that produces “assurance” (T 73, 124; cf. 106, EHU 27) in matters of fact that have not been observed. Does ‘assurance’ mean mere belief, a psychological state, or does it imply a state that involves knowledge, and hence carry epistemic implications? See my 2002, 61, 73, 77.

hume’s argument about induction

277

that no argument that is in fact sound causes such inference? Similarly, is his point that no probable argument that purports to be cogent causes inductive inference or that no genuinely cogent probable argument does so? On these questions, I agree with Peter Millican: “Hume focuses only on legitimate forms of argument, which on Garrett’s principles he has no right to do.” Millican was the first to advance a number of the points that follow.32 Hume’s intentions are clear in the case of demonstrative argument. He claims that there is no demonstrative reasoning to “prove” that nature is uniform (T 89, 92; EHU 35). This looks like a “success” verb; a proof is presumably a successful demonstration. Hume shows no interest in the question of whether inductive inference is caused by reasoning, however seductive, that falls short of a genuine proof of the uniformity principle. The descriptivist reading cannot explain Hume’s silence in regard to this possibility. Garrett replies that such arguments, “even if convincing, would not be instances of genuine demonstrations operating alone.”33 Granted, but were Hume merely pursuing cognitive psychology, he would need to exclude the possibility that inductive inference is caused by such impure reasoning. Bad arguments sway our thinking and practices all the time. Hume was certainly alert to purportedly demonstrative arguments that are flawed. After providing a general argument to show that there is no demonstration of the causal principle, he devotes five additional paragraphs to showing that its purported demonstrations—due to Hobbes, Clarke, Locke, and others—are “fallacious and sophistical” (T 80). All this is in I.iii.3. If the only issue in play in I.iii.6 is whether a demonstrative argument causes inductive inference, it is difficult to see why demonstrations that are subtly flawed, but have an air of plausibility, could not do the trick. I turn to the case of probable reasoning. A preliminary point. Why does Hume fail to consider inductive arguments that are not good ones? Garrett’s explanation is that even bad probable arguments rely on inductive extrapolation from the observed to the unobserved; for this reason, “they can no more be the underlying cause of inductive inference than ‘good’ or ‘philosophical’ ones can.”34 This line of defense depends on Garrett’s view that I.iii.6 seeks to explain the entire practice of inductive inference, rather than individual inductive inferences.35 This is dubious. It is custom or past experience that causes individual inductive inferences. Hume is clear in I.iii.16 that we cannot give “the ultimate reason, why past experience and observation produces such an effect” (T 179; cf. EHU 54–55). The ultimate reason for 32. Millican 1998, §VII, and 2002, 157–58. The quotation is from his 1998, 150. 33. Garrett 1998, 187. Garrett also finds “rather implausible” the hypothesis that inductive inference depends on “fallacious attempted demonstrations.” But Hume is alert to a wide range of psychological “distortions” both to belief, as in “unphilosophical probability,” and sentiment. See my 2002, ch. 4. 34. Garrett 1998, 187. 35. See Garrett 1997, esp. 91–92, and 2002, Appendix. Broughton contrasts her version of the position with that of Garrett and also begins to part company with the pure descriptivist position (2008).

278

reflection and the stability of belief

inductive inference, and hence the cause of the operation of custom itself, lies beyond Hume’s explanatory ambitions.36 Even bracketing this matter, the textual situation with regard to probable reasoning is more complex than in the case of demonstrative argument. Hume writes in the Treatise: “The same principle cannot be both the cause and effect of another” (T 90). This “causal” formulation perhaps opens the door to taking the issue to be confined to the causation of belief in the uniformity principle.37 This way of putting things, however, disappears in the Abstract: All probable arguments are built on the supposition, that there is this conformity betwixt the future and the past, and therefore can never prove it. . . . [O]ur experience in the past can be a proof of nothing for the future, but upon a supposition, that there is a resemblance betwixt them. (651–52) This reproduces language in the Treatise: “probability is founded on the presumption of a resemblance betwixt those objects, of which we have had experience, and those, of which we have had none; and therefore ’tis impossible this presumption can arise from probability” (90). The implicit point is that a probable argument for the uniformity of nature, the conformity of unobserved to observed, would be question-begging, presupposing the resemblance it seeks to establish. This is explicit in the Enquiry: “To endeavor . . . the proof [that the future will be conformable to the past] by probable arguments . . . must be evidently going in a circle, and taking that for granted, which is the very point in question” (EHU 35–36). Hume’s uptake on the causal formulation in the Treatise is significant: “The same principle cannot be both the cause and effect of another; and this is, perhaps, the only proposition concerning that relation [of causation], which is either intuitively or demonstratively certain” (T 90). Hume cannot resist the causal formulation, which affords the opportunity to inject a clever, but gratuitous, slap at reason.38 Taken literally, however, the formulation runs askew to the underlying point about circularity and thus can appear to support the descriptivist reading. This is ultimately “noise” in the text. Perhaps one could both brush off the Abstract and try to insist that Hume’s claims about probable arguments in the Treatise are merely psychological, whereas in the Enquiry they are epistemic. It is difficult to see any advantage in such maneuvers, since it is clear that Hume’s claims in the case of demonstrative arguments for the uniformity principle carry epistemic weight. Hume’s negative argument about reason manifests an epistemic dimension in the cases of both demonstration and probability. 36. For development of the point, see my 2004, pt. III. 37. As at Garrett 1997, 82, 91–92. 38. Bennett writes that the causal formulation “embodies a joke” (2001, 2.261).

hume’s argument about induction

279

5. Justified Inductive Inference as Hume’s Explanandum in I.iii.6 What does explain Hume’s failure to consider the possibility that reason supplies considerations that, though defective, are sufficiently appealing to cause us to make inductive inference? The answer must be that Hume imposes an epistemic constraint on any causal explanation of inductive inference: the explanation of our making inductive inferences must be compatible with their being justified.39 It seems plausible—plausible enough to attribute the thought to Hume—that if inductive inferences were caused by bad or defective reasoning, then they would not be justified, in violation of the constraint. This explains why Hume confines his attention to good or cogent considerations of reason—to genuine demonstrations and to probable reasoning that is not question-begging. Given that inductive inference is justified, it could not be caused by reason—bad reasoning is excluded, and there is no good reasoning to cause it.40 Attributing the epistemic constraint to Hume also accounts for his willingness to say, in the course of his main argument about induction itself, that “cause and effect . . . ’tis the only [connection or relation of objects], on which we can found a just inference from one object to another” (T 89). This is a statement of the data to be explained: the making of justified inductive inference. (The assumption, of course, is that some inductive inferences are justified, not that they all are.) The descriptivist reading obscures the epistemic structure of Hume’s argument. My suggestion does not reduce to the earlier point that Hume’s project in the science of human nature and his argumentative practices presuppose or assume that inductive inference is justified. This is a commonplace. As we have seen, even descriptivists must allow that Hume presupposes the justification of inductive inference. As Broughton puts things, there is a “natural presupposition,” one Hume shares, that inductive inference is reasonable.41 The linchpin in my interpretation is not that Hume’s engaging in inductive inference presupposes that it is justified, but that in I.iii.6 he incorporates this positive epistemic status into his description of the subject matter under investigation. Let me clarify the distinction here. Suppose Hume utilizes 39. This differs from Connon’s view that Hume takes inductive inference to be justified and seeks a description of what is involved in making inductive inference (1976, 129). Not any description would be compatible with justification. Criteria for justification are operating in the background. 40. We cannot regard the constraint as merely ad hominem against philosophers who maintain that reason does explain the belief in the uniformity principle. These friends of reason doubtless think that the arguments that figure in such explanations are good ones. Hume could defeat their position by showing that it is not the case that good reasons cause inductive inference. This would not, however, be sufficient to establish the position descriptivists attribute to Hume: that inductive inference is not due to reason. A showing that good reasons do not cause inductive inference leaves out of account the possibility that inductive inference is due to a defective employment of reason, rather than to custom. We must therefore suppose that the assumption that inductive inference is justified is Hume’s own. 41. Broughton 1983, 5, 12. See also Garrett 1998, 186.

280

reflection and the stability of belief

inductive inference in the course of the I.iii.6 investigation into the causes of inductive inference. This would not itself explain the role of epistemic considerations in the section. In order to explain this, we would have to take the assumption that inductive inference is justified to be operative at two levels: in the method of investigation, but also in Hume’s understanding of the phenomenon he seeks to explain. Another way to see the distinction is as follows. It is clear that Hume relies on inductive inference in I.iii.8, where he argues from “experience and observation” (T 101), from “experiments” (99, 100, 102), that causal inference is due to custom or habit. By contrast, the I.iii.6 discussion of the uniformity principle does not utilize inductive inference. Rather, Hume is relying on philosophical reflection—considerations about the conceivability of the falsehood of the uniformity principle, considerations about the presuppositions of any probable argument—to reveal limitations in demonstrative and probable arguments. In the core of I.iii.6 (paragraphs four through eleven), the assumption that inductive inference is justified is part and parcel of the phenomenon under investigation, but not integral to the method of investigation itself. The main argument in I.iii.6 is shot through with epistemological claims and can hardly be regarded merely as an exercise in descriptive psychology.

6. Hume’s Opposition to Epistemic Parasitism A closely related textual consideration arises in connection with I.iii.8. Recall that a person who stops his journey at a river “foresees the consequences of his proceeding forward.” Hume adds some important comments: “custom operates before we have time for reflection,” without “a moment’s delay”; we draw the inference from past experience “without reflecting on it” (T 104, emphasis added). The riverbank passage poses a challenge to the descriptivist reading: if in I.iii.6 Hume is merely making the negative point that inductive inference is not caused by reason, why does he not rest content with the empirical observation that reflection and reason play no role in typical cases of basic inductive inference? Indeed, Hume does make this point in I.iii.6: “the imagination of itself supplies the place of . . . reflection” and does so without “a moment’s delay” (93). This simple observation would seem to dispose of reason as the causal explanation of inductive inference.42 If so, how is the descriptivist to explain the central position of the considerations about demonstrative and probable arguments in the section? An argument in “Of skepticism with regard to the senses” heightens the puzzle. Hume asks “whether it be the senses, reason, or the imagination, that produces” (T 188) the belief in continued and independent objects. This looks a lot like asking, as Hume does in I.iii.6 (88–89, 92), whether it is 42. See Millican 1998, 152, and Broughton 2008, 297–98.

hume’s argument about induction

281

reason or the imagination that produces inductive inference. Here is Hume’s case against reason in I.iv.2: “whatever convincing arguments philosophers may fancy they can produce to establish the belief of objects independent of the mind, ’tis obvious these arguments are known but to very few, and that ’tis not by them, that children, peasants, and the greatest part of mankind are induc’d” (193) to believe in body. Hume relies on the empirical observation that most of us believe in body without interposing any reflection or reason. If that is a good argument against reason’s causing the belief in body, why do not analogous considerations—the person at the riverbank—suffice to show that reason does not cause inductive inference? Why cannot Hume rest content to observe that basic inductive inferences proceed without a moment’s delay, that they are not caused by reflection or reason, and a fortiori they are not caused by any good or cogent deployment of reason? The answer is that he wants to foreclose a broad strategy for explaining why inductive inference is justified. Suppose that reason, fully deployed, could produce cogent reflections in favor of the uniformity principle. In that event, friends of reason—those who think that the justification of inductive inference somehow resides in reason—might propose a position along the following lines. Even if the person at the riverbank does not himself invoke the cogent reasoning, even if he is not so much as aware of it, his inductive inferences might nevertheless possess justification in virtue of the availability of the cogent argument; such justification would be indirect, parasitic upon the existence of reasoning available to a more reflective person. Let us call this strategy epistemic parasitism. How might indirect justification be thought to arise? Consider Descartes. We attain unshakable certainty, scientific knowledge, that the material world exists only if we first prove the existence of a nondeceiving God. Where does that leave those of us who have not proved this, and hence do not possess certain or scientific—let’s say lofty—knowledge? Evidently, such knowledge of the material world as you and I possess is second-rate. One way to flesh this out is to say that, even though we have not proved a nondeceiving God, we do possess a fragment of the system of beliefs required for lofty knowledge. Perhaps our sketchy belief system confers on us second-rate knowledge. Alternatively, perhaps the fact that we could obtain lofty knowledge, were we fully to exercise our faculty of reason, confers second-rate knowledge. Either way, our lowly knowledge derives from the availability of a cogent argument for a nondeceiving God.43 Epistemic parasitism cannot be defeated by the observation that ordinary cognizers—as at the riverbank—do not reason or reflect; the whole point of parasitism is that justification for perfectly ordinary persons is derivative from the availability of lofty knowledge that they do not themselves possess. Hume wants to preempt parasitism by showing that putative lofty knowledge is unattainable in the first place, so that no other knowledge can be second-rate. In connection with inductive inference, Hume does take himself 43. I say more about these themes in my 2008, 115–17, and 2004, §§II.1, IV.4.

282

reflection and the stability of belief

to show that lofty knowledge is unattainable; there is no (demonstrative or probable) argument available to reason that justifies belief in the uniformity principle. We thus cannot regard the argument about induction in I.iii.6 as a descriptive exercise in cognitive psychology. Hume wants to quash overly reflectivist accounts of the way the mind ordinarily works, but he also needs to take on overly intellectualist accounts of the way the mind could work even in the case of an ideal, fully reflective cognizer.44 To this end, the riverbank argument can at best play a role that is supplementary to the main argument about induction.45 What about Treatise I.iv.2, where Hume does employ an empirical argument: “children, peasants, and the greatest part of mankind” have no knowledge of any considerations of reason that support the belief in body? In this context, there is no harm in employing the short argument. In Part iii, Hume’s claim that reason could not cause us to make causal inferences, given that such inferences are justified, is accompanied by the thesis that all assurance in unobserved matters of fact is based on causal inference (T 73–74, 87, 89, 107–8) and the identification of “reason” with causal inference (117–18n.). Both the terminological stipulation (193; cf. 231) and the substantive thesis about the scope of causal inference (193, 212; cf. 198, 216) are imported into I.iv.2. Part iii has destroyed parasitism as an account of knowledge of any unobserved matters of fact. In Part iv, the earlier care in addressing parasitism is no longer required. We have seen that a generalized descriptivist interpretation faces numerous difficulties. Hume persistently provides positive assessments of inductive inference, and does so without comment or qualification. These assessments include evaluations of second-order inductive principles. There is the continuity in his posture toward inductive inferences across Part iii and much of Part iv of Book I, and also a recurring normative distinction between reason and the imagination. In addition, the riverbank passage in I.iii.8, together with the character of Hume’s main argument about induction, poses severe difficulties for the descriptivist account of I.iii.6. Even the most casual reader will be struck by the pronounced epistemic flavor of the section.

7. Hume’s Epistemological Externalism Does recognition of the epistemic character of I.iii.6 saddle us with the skeptical interpretation of the section after all? It would, if Hume thought that a belief is justified only if it is supported or supportable by good argument. Roughly speaking, this is an internalist assumption. According to epistemological internalism, the justification of a belief depends exclusively on the 44. Broughton emphasizes the first of the two ambitions (2008). 45. Cf. Millican 1998, 151–52. Hume concludes the riverbank passage: “This removes all pretext, if there yet remains any, for asserting that the mind is convinc’d by reasoning” (T 104, emphasis added) of the uniformity principle.

hume’s argument about induction

283

beliefs one holds or—in versions that tolerate parasitism—on beliefs that are available, beliefs one could hold. The Russellian “problem of induction” relies on internalist presuppositions in moving from the claim that there is no cogent (demonstrative or probable) argument for the uniformity of nature to the conclusion that no inductive belief has any justification. Even supposing that demonstrative and probable arguments exhaust the field, the Russellian skeptic requires the premise that an inductive belief can be justified only by argument, argument taking as premises other beliefs one holds or could hold.46 In light of the massive evidence that Hume is not a skeptic about induction, he must reject this internalist way of thinking. Hume does contend that there are no cogent arguments for the uniformity of nature. Yet, he labels inductive inference “just” in the course of his main argument about induction and maintains his commitment to inductive inference in the remainder of Part iii and at least much of Part iv. This is unaccountable, on the supposition that Hume is an internalist.47 Hume contributes key premises for the skeptical argument about induction, but he does not share the internalist framework that is also necessary to generate its conclusion. In order to avoid attributing skepticism to Hume, we need not suppose that his argument in I.iii.6 is only directed against a specialized conception of “reason” (§3). Skepticism can be avoided, even if there is no good argument of any sort supporting the uniformity principle. Let us see more concretely how Hume can stand by his positive epistemic assessment of inductive inference, though such inference is not justified in virtue of argument. Hume’s negative claim in I.iii.6, that inductive inference is not due to reason, leads to the positive claim in I.iii.8, that it is due to custom. In I.iii.12, “Of the probability of causes,” Hume extends the role of custom to the explanation of forms of probabilistic or statistical inference that are “receiv’d by philosophers” (T 143). Once reason or argument has been ruled out, custom is the only resource Hume has available to account for induction’s positive epistemic status. I do not mean that inductive inference is justified simply because it is due to custom or repetition. Hume stresses in I.iii.9 that education or indoctrination is also due to custom but is “disclaim’d by philosophy, as a fallacious ground of assent”(118).48 At the same time, if inductive inference is justified, but not by argument, custom is the only place to look for its justification. There must be some feature of custom, as it operates in inductive inference, that generates justification.

46. Garrett notes that the skeptical reading requires attributing this premise to Hume and observes that he nowhere advances it (1997, 82–83). As we shall see, Hume rejects the premise. Indeed, it is rejected implicitly as early as I.iii.6. 47. If he is an internalist, as characterized here. There is room to attribute to Hume a “negative” coherence theory of justification. See my 2001b [this volume, ch. 5]. 48. Whether Hume can sustain this epistemic distinction between two kinds of custom is a difficult matter central to his epistemological project. See my 2002, §VII.5.

284

reflection and the stability of belief

In pioneering work, Kemp Smith in effect identified the irresistibility and unavoidability of custom as the crucial feature.49 A number of other features of custom, however, suggest themselves as candidates for the central role in Hume’s epistemology. Perhaps the operation of custom confers justification insofar as it is an operation of a healthy organism, or because it is adaptive, or because it is reliable. These options give rise to proper function, adaptivist, and reliabilist interpretations of Hume.50 My own view is that what matters for Hume is custom’s tendency to infix ideas, to produce steady beliefs.51 Adjudicating among these alternatives—better yet, understanding how the texts and Hume’s historical position constrain the choice among them—is the central challenge in coming to terms with Hume’s epistemological position.52 What matters here is that the various options—appealing to irresistibility, proper functioning, adaptiveness, reliability, stability—are externalist theories. According to these theories, the epistemic status of a belief depends, at least in part, on naturalistic facts about the mechanism that produces it. Hume is offering an externalist account of why inductive inference is justified. In a recent book on the problem of induction, reliabilism is the first solution considered.53 Russell, an internalist and foundationalist, had no such solution remotely in view. Attributing externalism to Hume explains how he could reject internalism and thus avoid drawing a skeptical conclusion at the close of I.iii.6. It also explains why he cannot rely on the riverbank considerations, at least insofar as Hume wants to argue against internalism; they do not defeat parasitism as an internalist strategy. An externalist interpretation has considerable explanatory power and coheres with Hume’s focus on custom as the mechanism that underpins inductive inference. Of course, we would want to locate direct textual evidence for Hume’s externalism. I cannot construct the case here, but I can show that there is such evidence. The opening paragraph of I.iv.4 is one central place to look. A person who infers from the articulate voice that someone is near “reasons justly and naturally.” By contrast: [O]ne, who is tormented he knows not why, with the apprehension of spectres in the dark, may, perhaps, be said to reason, and to reason naturally too: But then it must be in the same sense, that a malady is said to be natural; as arising from natural causes, tho’ it be contrary to health. (T 225–26) This resonates with adaptivist theories. So, too, does I.iii.10, where “Nature has . . . chosen a medium” between perceptions, on the one hand, and the 49. Kemp Smith 1905 and 1941. See my 2002, 20–25, for difficulties. 50. For a proper function interpretation, see Craig 1987, 81, and Wolterstorff 1996, 166n.6; for adaptivism, Schmitt 1992, 68–72; for reliabilism, Schmitt 1992 and 2004. 51. I advance this view in my 2002. I amend my position in 2004. 52. For a start on this project, see my 2004, pt. III. 53. Howson 2000.

hume’s argument about induction

285

“unsteadiness” of mere ideas, on the other, so that we might avoid future “goods and evils” (119). The medium is belief, and steadiness in the service of guiding action is its natural function. This passage also resonates with proper function and stability theories. Section I.iii.16 is germane as well to externalism in the Treatise. In discussing the riverbank passage, I suggested that Hume seeks to thwart epistemic parasitism. His commitment to externalism lies at the root of this objective. The pressure toward parasitism arises within an internalist framework. From an internalist perspective, the arguments required to justify fundamental beliefs—the uniformity of nature, the existence of the external world, and so forth—must meet some threshold of articulation and sophistication. Suppose the internalist sets the threshold low. Even so, it is implausible that “honest gentlemen” (T 272) or landholders, peasants (193), and the vulgar at large, as well as children (193), reach the required threshold. The internalist must resort to providing, in the first instance, an account of the justification of suitably idealized cognizers.54 Justification and knowledge for the vulgar and children is derivative from that of idealized cognizers. Within externalism, this pressure toward parasitism vanishes. Custom operates in the vulgar and in children; the favorable epistemic status of their inductive beliefs can be explained in terms of the relevant features of custom, without any reference to belief systems available to more reflective persons. Indeed, it is part of Hume’s enterprise to secure the epistemic standing of many beliefs in nonhuman animals. This is explicit in the Enquiry: “the most ignorant and stupid peasants, nay infants, nay even brute beasts, improve by experience, and learn the qualities of natural objects, by observing the effects, which result from them” (EHU 39, emphasis added; cf. 105). The externalist theories provide a direct explanation of the epistemic achievements of nonhuman animals. The Treatise, as well as the Enquiry, devotes a section to the cognitive achievements, the “reason,” of nonhuman animals. Custom operates “all the way down,” from the most reflective philosophers to nonhuman animals. Hume’s appeal to custom is thus central to his constructive project of explaining how beliefs of all these organisms can constitute knowledge in their own right. Epistemic parasitism is antithetical to Hume’s competing picture, in which the most ordinary creatures possess full-fledged knowledge. We might summarize some of our results as follows. Section I.iii.6 can be neither purely descriptive nor purely skeptical in intent. Not purely descriptive— that inductive inference is justified is built into the characterization of the phenomenon to be explained and constrains the admissible explanations. And not purely skeptical—Hume’s endorsement of induction survives both the negative claim that inductive inference is not caused by reason and the constructive claim that it is caused by custom. This does not leave Hume 54. This formulation applies both for foundationalist and coherentist versions of internalism. I leave open the extent to which the latter are in play in Hume. See my 2008, 115–17.

286

reflection and the stability of belief

and his interpreters in the position of having to walk a fine line. To the contrary, externalism represents a broad swath of middle ground, allowing Hume to be engaged in normative epistemology without embracing the Russellian problem of induction. The difficulty has been to get this territory into commentators’ field of vision. As Garrett observes, Hume relies on inductive inference “before, during, and after” I.iii.6.55 We can suppose that Hume just could not help himself, or we can recognize that his commitment to induction is sustained by an externalist orientation in epistemology, and hence is impervious to internalist or Russellian attack. At this juncture, we are positioned to entertain a way of thinking about the motivation for descriptivism. Consider the old saw that Hume confuses psychology with philosophy. On its face, this is uncharitable. We might think of proponents of descriptivism as opting for a more charitable version of this line: Hume does not confuse psychology with philosophy; he selfconsciously separates them. As we have seen, this is not Hume. To the contrary, it requires us to jettison one of his most distinctive philosophical contributions: that the psychology of belief formation, the characteristics of the psychological mechanisms that produce belief, are critical determinants of the epistemic status of belief. Who would have thought Hume’s view a confusion? The answer, painting with a broad brush, is interpreters under the sway of logical empiricism, where we find the claim that philosophy, and hence epistemology, is a priori. Psychology is at best relevant to “the context of discovery,” not to “the context of justification.” Theories of evidence are essentially theories of a priori confirmation relations. For Hume to think that evidence depends on genetic facts about belief then appears to be a confusion. This would have bothered positivistically inclined interpreters. After all, logical positivism claimed Hume as one of its own insofar as he stressed the divide between “relations of ideas” and “matters of fact,” in the service of an onslaught on metaphysics. Interpreters in the positivist tradition thus strived to save Hume from himself. The positivist tradition, however, predates more naturalistic strands in twentieth-century epistemology. Positivists were either foundationalists (as with Schlick) or coherentists (as with Neurath), but internalists either way. They did not contemplate, and certainly did not accept, externalist theories. We can think of descriptivism as resulting from damage control within an interpretive tradition that does not see externalism as an option. I am unsure that in aligning himself with externalism Hume was ahead of his time, but he was certainly ahead of his positivistically inclined twentieth-century interpreters.56 55. Garrett 1997, 78. 56. I am grateful to Don Garrett and the other participants in the April 2005 Wake Forest University Hester Conference on Hume’s Naturalism for lively and useful discussion. I am indebted especially to Janet Broughton and Peter Millican for helping me advance my thinking on the topic of this chapter.

hume’s argument about induction

287

11 Locke and British Empiricism

1. Lockean Apparatus in part i of Book I of Hume’s A Treatise of Human Nature, we encounter divisions between simple and complex ideas (T 2, 4), sensation and reflection as sources of ideas (7–8), and complex ideas of relations, modes, and substances (13).1 These Lockean distinctions, familiar from An Essay concerning Human Understanding II.i, II.ii, and II.xii, find themselves in the company of Lockean doctrines. There is a succinct reformulation and endorsement of Locke’s denial of innate ideas (7) and, by implication, his denial of innate knowledge. For Hume, all complex ideas are formed from simple ideas that correspond to impressions (3–5). That all ideas derive from experience is the overarching thesis of Essay Book II. According to Locke, the idea of a substratum or “support” is at best obscure (negative and indistinct) and relative (Essay I.iv.18, II.xii.6, II.xiii.17–20, II.xxii.1–2; Works 4.21). Hume argues that there is no idea of substance, construed as “an unknown something, in which [particular qualities] are supposed to inhere,” derived from sensation or reflection, and hence that ‘substance’, distinct from a collection of qualities, has no “meaning” (T 16). The argument presupposes that meaningful words stand for ideas, the central claim of Essay Book III. A Lockean reader would feel at home, though perhaps a bit breathless, as Hume concentrates essentials of Books I, II, and III of the Essay into a few short sections. In addition, the Treatise I.i section on the association of ideas 1. [In this chapter, quotations of Hume are based on the Selby-Bigge and Nidditch editions of the Treatise and Enquiries. In-text notes in the original article are here converted to footnotes.]

288

is reminiscent of the final section of Book II of the Essay in the fourth and fifth editions. All this is within the first sixteen or seventeen (Selby-Bigge) pages of the Treatise. In themselves, these similarities tell us little. Our knowledge of the composition of the Treatise is slight.2 We have two complete early drafts of Locke’s Essay, no drafts of the Treatise. Hume might have composed I.i before the remainder of the Book, and not revised in light of subsequent departures from Locke’s views. Or perhaps Hume wanted the opening sections of the Treatise to pay homage to Locke, lulling a reader into a false sense of security in advance of innovations—ones Lockeans might not find congenial. Michael Ayers suggests that the first section of Part I of Berkeley’s Principles of Human Knowledge has just such an artful construction.3 Or Hume might not have been clear about where Lockean principles give way to more distinctively Humean arguments. In any case, Locke’s programmatic views about meaning and concept acquisition do not dictate specific applications. Locke thought that the clearest idea of active power derives from observing the mind’s command over its ideas and limbs (Essay II.xxi.4–5); observing the transfer of motion in impact also gives us an idea—albeit “a very imperfect obscure Idea”—of active power (II.xxi.4). Berkeley denied this latter claim: the (related) idea of causation is derived exclusively from the experience of willing ideas, of volitional activity (PHK 25, 27–28; DHP II, 217, III, 231); the concept of causality has no legitimate extension beyond spirits and their volitions. Hume thought the idea of causation derives from the experience of constant conjunction, perhaps together with that of a mental determination to expect an object’s usual attendant. The malleability of empiricist theories of meaning, whether in the eighteenth or twentieth century, enables Berkeley and Hume to advance opposing, empiricist theories of causation.4

2. Demonstrative Knowledge Before returning to Hume’s meaning empiricism in §5, I consider Locke’s relationship to Hume in other respects. Though an empiricist about the origin of concepts, Locke accepted demonstrative (and what was later to be called “a priori”) knowledge of propositions, even outside logic and mathematics. Demonstrative knowledge is possible where “nominal” and “real” essences coincide (Essay III.v.14, 16; cf. II.xxxi.6). Locke is confident that the nominal and real essences of ethical concepts can coincide (Essay III.xi.16), that moral truths admit of demonstration (I.iii.1, III.xi.15–18, IV.iii.18–20, IV.iv.7–9, IV.xii.8, 11). Demonstrative knowledge is often “instructive,” rather than “trifling,” where we have “but a verbal Certainty” (IV.viii.8). 2. See Norton and Norton 2007, 434ff. 3. Ayers 2005, 47–48. 4. For the malleability in twentieth-century theories of meaning, see Hempel 1965.

locke and british empiricism

289

The trifling-instructive distinction is related to Hume’s “relations of ideas” (propositions, given the Lockean identification of meaning with ideas, true by virtue of meaning) and “matters of fact” (cf. EHU 25), and to the analyticsynthetic distinction within logical empiricism. Hume and his positivistic successors, however, insisted that only analytic propositions are necessarily true and knowable a priori (25–26). Hume often targets rationalist claims to a priori knowledge on this ground. Here we put to the side Hume’s Treatise III.i.1 argument, against Locke and other moral rationalists, that “moral distinctions” are derived from sentiment, not reason. Locke can seem to suggest that it is in principle possible to acquire demonstrative knowledge of some laws of nature, if only we knew the real essences of bodies (Essay IV.iii.16, 25; cf. IV.vi.10–11). For Hume, it is conceivable, and hence possible, that particular uniformities fail to hold (T 86, 91–92, 161–62; EHU 28–30, 32–34); there are no “necessary Connexion[s]” or “necessary co-existence[s]” (Essay IV.iii.16, IV.vi.10) among the qualities of bodies for reason to uncover. Locke also held that we have intuitive knowledge that an effect’s properties “must be owing to, and received from” (Essay IV.x.4) its cause. This constraint on causality is stronger than Descartes’ principle that a cause must contain at least as much (overall) perfection as its effect (CSM 2.28: AT 7.40). Locke requires that the cause must contain at least as much of each perfection in the effect—at least as much knowledge, power, and so forth (Essay IV.x.4–6, 12). Hume rejected all such restrictions on possible causal relations: “to consider the matter a priori, any thing may produce any thing” (T 247). Convinced that the laws of nature cannot be demonstrated, and that intuitive knowledge of the metaphysics of causation and demonstrative knowledge of God and morality are in principle beyond our reach, Hume found demonstrative knowledge of little theoretical interest. Though these were much more than intramural differences, Locke’s most far-reaching influence on Hume came from another quarter entirely.

3. Hume and Locke’s Restrictions on the Third Degree of Knowledge After claiming at Essay IV.xi.1–8 that we have “sensitive knowledge” of the existence of external objects, Locke calls attention to a limitation: [W]hen our Senses do actually convey into our Understandings any Idea, we cannot but be satisfied, that there doth something at that time really exist without us. . . . But this Knowledge extends as far as the present Testimony of our Senses . . . and no farther. For if I saw such a Collection of simple Ideas, as is wont to be called Man, existing together one minute since, and am now alone, I cannot be certain, that the same Man exists now. (Essay IV.xi.9; cf. IV.iii.5, 21)

290

reflection and the stability of belief

Two sections following, Locke liberalizes this account: “so by our Memory we may be assured, that heretofore Things, that affected our Senses, have existed. . . . But this Knowledge also reaches no farther than our Senses have formerly assured us” (IV.xi.11). Even allowing sensitive knowledge mediated by memory, knowledge of the external world remains extraordinarily fragmentary. If I see Locke’s “Man” every day, but only at noon, I do not know that he has existed or will exist at any other time, past or future. Locke did not himself consider this result problematic. He allows that “it be highly probable that Millions of Men do now exist,” though “I have not that Certainty of it which, we strictly call Knowledge” (IV.xi.9). He intends his IV.xv–xvi account of probable belief, backed by “Constancy of experience” (IV.xi.16), to support this assessment. Hume found the lacuna at IV.xi.9–11 disconcerting and sought to repair it. There is striking evidence for this in the first Enquiry (hereinafter, “the Enquiry”): “It may . . . be a subject worthy of curiosity, to enquire what is the nature of that evidence which assures us of any real existence and matter of fact, beyond the present testimony of our senses, or the records of our memory” (EHU 26). “[T]he present testimony of our senses” is also Locke’s phrase at IV.xi.9. Hume seeks to identify the evidence that enables us to extend our assurance beyond its Lockean boundary. Hume formulates his answer straightaway: “By means of [the relation of Cause and Effect] alone we can go beyond the evidence of our memory and senses” (26). Locke’s texts set the problem. They also might have suggested to Hume his solution. Locke maintains that he “cannot be certain, that the same Man exists now,” on the ground that “there is no necessary connexion of his Existence a minute since, with his Existence now” (Essay IV.xi.9, emphasis added). Hume’s insight is that there is a causal connection, whatever sort of necessity that involves. Even if causal connections are not necessary truths susceptible to demonstration, one can infer the man’s existence at times he is not observed as the unobserved cause (or effect) of his existence at times he is observed. There has long been a strong case that Hume is not a skeptic about inductive inference, contrary to the dominant interpretation from the 1940s into the 1970s.5 Throughout I.iii, Hume characterizes causal inference in terms suggestive of an epistemic achievement: the relation of cause and effect “informs us of existences and objects, which we do not see or feel” (T 74, emphasis added; cf. 73), “brings us acquainted with . . . existences . . . beyond the reach of the senses and memory” (108, emphasis added; cf. 103). Hume insists, against Locke, that “many arguments from causation exceed probability” and constitute “proofs” (124; cf. EHU 164). Hume’s is a causal theory of knowledge (cf. EHU 55). G. E. Moore spotlighted Hume’s account and lavished considerable attention on it in his 1910–11 lectures.6 Hume

5. For the case against the skeptical reading, see Beauchamp and Mappes 1975; Connon 1976; Winters 1979; Beauchamp and Rosenberg 1981; and Broughton 1983. 6. See G. E. Moore 1909 and 1953, respectively.

locke and british empiricism

291

supplements sensitive knowledge with an elegant theory of knowledge of the unobserved.

4. Association and Custom In the Treatise, Hume devotes a scant few pages—the opening section of I.iii— to demonstrative knowledge. The causal theory of knowledge is introduced in I.iii.2 (T 73–74; cf. 82, 84). In this context, Hume examines another Lockean claim to certainty (Essay IV.x.3), that we have intuitive knowledge that whatever has a beginning must have a cause (T 78). Hume’s response to Locke’s account of sensitive knowledge thus jump-starts his further distinctive contributions in I.iii: his account of causal necessity at I.iii.14, where Hume deems Locke’s brisk analysis of causation (Essay II.xxvi.1) too thin to be helpful (T 157); and his associationist account of causal inference at I.iii.6–13. Here Locke also influenced Hume, again in an oblique way. Hume had access to the fifth edition of the Essay.7 Locke had introduced the section on associationism in the fourth. For Locke, association plays a subsidiary role,8 accounting for “Unreasonableness” (Essay II.xxxiii.1, 3), opinion in “opposition to Reason” (II.xxxiii.4).9 Associative connections are “undue” or “wrong” (II.xxxiii.8, 9, 18), requiring “remedies” (Works 3.276). They are “accidental” (Essay II.xxxiii.7), due to “Chance or Custom” (II.xxxiii.5); “Custom settles habits of Thinking in the understanding,” resulting in “an habitual train” (II.xxxiii.6) of ideas. Hobbes had treated causal inference associationistically (Lev. iii, 7; EL I, IV, 7), but Hume’s reading of Locke likely suggested an explanatory role for repetition. Hume sees that enumerative induction inherently implicates custom (cf. T 102)—repeated observation of similar pairs of objects, giving rise to habits (cf. 128, 153) of association and “settled” (108, 608) or “infixed” (cf. 86, 225, 453) belief in the unobserved. For Locke, custom settles haphazard associations; in Hume’s hands, it cements reasonable habits of inference.10 We can credit Hume both with recognizing that causal connections are necessity enough to sustain knowledge of the unobserved and with appropriating custom to explain associative transitions in causal inference. This is the story of creative breakthroughs and transformations against the background of Lockean materials. In seeking to explain (nearly) all doxastic phenomena associationistically (cf. T 8–13; Abs. 661–62), Hume departed from both Hobbes and Locke.11 Empirical investigation now extends to the cognitive faculties themselves.

7. Norton and Norton 2007, 980, 1019. 8. Not so much as meriting a chapter in the thirty-plus article volume, A Companion to Locke, for which this essay was commissioned. 9. J. Gibson 1917, 234–36. 10. Aaron 1937, 141. 11. Laird 1932, 39–41; Yolton 1993, 18–23; and my 1997.

292

reflection and the stability of belief

Hume’s constructive epistemological project proceeds within this framework. Whereas for Locke intuition and demonstration uncover “probable” as well as “necessary connexion[s]” (Essay IV.xvii.2), Hume provides a thoroughgoing naturalistic account of probable reasoning. Bracketing demonstration, Hume identifies “reason” with “those conclusions we form from cause and effect” (T 231; cf. 193). “[T]he customary transition from causes to effects, and from effect to causes” is due to “permanent, irresistable, and universal” (225) associative principles. Unreasonable beliefs, “whimsies and prejudices” (118n.), are due to associative propensities other than causal inference. A person apprehensive of specters in the dark is in the grip of principles that are “changeable, weak, and irregular” (225–26); one who, hearing an articulate voice in the dark, infers, by custom, that someone is present, “reasons justly” (cf. 89, 144).

5. Hume’s Meaning Empiricism Reconsidered Empiricism about meaning did have a deep impact on Hume: we have no idea of a vacuum (T 53–64), time without change (36–37, 64–65), external existence (67–68), necessary connection—distinct from a felt determination of the mind—(165–66), or substratum, whether material (16) or immaterial (232–34, 251–52). At the same time, such inventories provide a one-sided picture.12 The semantic arguments are often perfunctory, superseded by multistage psychological explanations of the very beliefs one could not possess, were the destructive meaning empiricist arguments correct. Think of some melting wax. (1) When we gradually survey its sensible qualities, we attribute to them a strict identity, as though they were unchanging as well as uninterrupted (T 60–62). (2) When we compare the qualities of the wax before and after they undergo considerable change, from a hard cube to a molten mass, we recognize that we are not viewing an identical object. (3) The two ways of considering the qualities lead to “a contrariety in our method of thinking” (220). (4) The response is to suppose a substratum as a locus of strict identity. Whereas Hume explains stage (1) associationistically, the transition from (3) to (4) outruns even his generalized associationism: (4) puts philosophers “at ease” (T 224; cf. 205, 215), alleviating the conflict at stage (3).13 Hume read “Of Power” (157n.) in the fifth edition, after Locke had introduced uneasiness to explain the determination of the will (Essay II.xxi.29–47). Locke’s text suggests a more general doctrine: “The motive, for continuing in the same State or Action, is only the present satisfaction in it; The motive to change, is always some uneasiness” (II.xxi.29). Berkeley rejected this motivational story,14 perhaps owing to its seemingly deterministic character. 12. See my 2001a [this volume, ch. 6]. 13. For the point that Hume does not confine himself to associationist mechanisms, cf. Passmore 1952/1968, 122. 14. Luce 1944, 352.

locke and british empiricism

293

I suggest Hume applied it to belief (cf. T 447, 453, 625–26). Yet another innovative adaptation of Lockean resources, though focused on relieving uneasiness arising specifically from conflicting beliefs—a distinctive psychodynamic twist. The explanation of the ancients’ belief in material substratum is not “mere psychology.” At stage (1), we are inclined to attribute identity to observable qualities. The belief in an unobservable substratum at stage (4) is “the means by which we endeavour to conceal” (T 219, emphases altered) the contradiction at stage (3). This superficially masks a conflict (cf. 254) that arises only because the gradual survey at stage (1) “deceives the mind” (220); owing to a “trivial propensity of the imagination” (224), we mistake diversity for identity. Causal inference plays no role. The next paragraph, the opening paragraph of I.iv.4, contrasts causal inference with the “changeable, weak, and irregular” mechanisms—precisely to make clear why Hume is not “unjust in blaming the ancient philosophers for making use of [the imagination]” (225). Hume’s explanations of belief in immaterial substrata, external existence, and necessary connection (167, 211–16, 253–55), follow a similar pattern; they are, one and all, epistemically debunking.15

6. Meaning Empiricism and Cartesian Metaphysics in Berkeley What is the role of empiricism about meaning in Berkeley? The bishop’s philosophical lineages, and their interactions, are complex matters. Studies dating from the 1930s show that Malebranche had a significant influence on Berkeley’s thought.16 It is hard to deny that Berkeley subscribed to a Cartesian metaphysics of mind.17 For Locke, the relative ideas of material and immaterial substrata are “equally obscure, or none at all” (Essay II.xxiii.15; cf. II.xxiii.27). Berkeley agreed with respect to material substratum (PHK 16–17, 68; DHP III, 198–99) but saw “no parity of case” (DHP III, 234; cf. 250) with souls. Ideas “inhere in” or are “supported by” spirits in that they are perceived by them (PHK 2, 27, 89, 135, 139; DHP III, 231–34, 237); the metaphors can be cashed out. This rolls back Locke’s chief application of his meaning empiricism. For Berkeley, as for Descartes (CSM 1.196: AT 9B.8), mental states must necessarily inhere in a substance (PHK 3, 33; DHP II, 206, III, 234). Because souls are simple, they cannot be destroyed by decomposition; they are “naturally immortal” (PHK 141). Since for spirits, to be is to perceive (PHK 139), or perhaps to perceive or to will (cf. PN 429, 646), a mind cannot exist unless it has conscious states. This is another Cartesian doctrine. In the

15. See my 2002, ch. 5; cf. Craig 1987 and Pears 1990. 16. Aaron 1931; Hicks 1932, 229–36; Luce 1934 and 1963; and Jessop 1938. 17. Bracken 1974 and McCracken 1988.

294

reflection and the stability of belief

Dialogues, Berkeley allows “immediate knowledge” of one’s own mind (DHP III, 232; cf. 231). This goes beyond Descartes, who held that minds can be known only through their accidents (CSM 1.210, 2.124, 156: AT 9B.25, 7.176, 222), though what each figure held is controversial. Berkeley’s Cartesian theory of mind is set within an occasionalist metaphysics; God—subject to some strategic exceptions (PHK 116, 147; DHP III, 237)—is the direct cause of successive ideas.18 Had Berkeley merely purged Locke’s system of mind-independent bodies, perceptual experiences would exist without a cause. For Malebranche, God is the sole cause of both physical and mental states (with an exception for desires—in the Search after Truth 1.1.2/4–5, 4.1.3/266–67 and Elucidations 1 and 2 at LO 552–55 and 559–60). Bodies exist but are causally idle. Removing body from Malebranche’s system results in Berkeley’s basic occasionalist structure,19 despite factoring out significant doctrines, the “intelligible extension” and “seeing all things in God” (PHK 148; DHP II, 213–14). There is the curiosity that Berkeley’s empiricist analysis of causation provides crucial premises—that sensory experiences cannot be caused by one another (PHK 25, 64, 102; DHP III, 231), nor by matter, did matter exist (PHK 68)—for an eliminative and demonstrative argument that they are caused by another spirit. Similarly, consider the metaphysical problem at Principles §45. If for sensible things, to be is to be perceived, is the table annihilated and created anew whenever one leaves and returns to the study? Berkeley entertains phenomenalism: to say the table exists means that if one were suitably situated, one would have “table-ish” sensory experiences (PHK 3, 58; cf. DHP III, 253)—an analysis developed by the logical empiricists, who identified statement meaning with methods of experiential verification.20 Berkeley floats other responses, most notably that “full-bodied” sensible objects—an entire table or apple, including its sensible interiors, over time—exist as perceptions (DHP II, 211–12) or “archetypes” (DHP III, 240, 248, 254) in the Divine mind. Another response is relative indifference to the problem (cf. PHK 45). Pressing meaning empiricism in the phenomenalist direction is but one option in view. Berkeley faces a structurally similar issue about the continuity of minds. If the mind is essentially conscious, does it cease to exist during deep sleep or comas? Descartes responds that these states impair memory, not consciousness itself (CSM 2.171–72, 246–47: AT 7.246, 356–57). Berkeley’s response pushes hard on a Lockean thread: “we have our notion of Succession and Duration from . . . Reflection on the train of Ideas . . . in our own minds” (Essay II.xiv.4). For Berkeley, time is nothing but a succession of ideas in a mind; absent conscious states, no time passes, so that there is no discontinuity 18. Cf. Jolley 1990. 19. See my 1981a, ch. 6; McCracken 1983, ch. 6; and Jolley 1996. 20. In “Berkeley’s Life and Writings,” John Stuart Mill, citing Principles §3, focused more on the element of potentiality than on the semantic claim (CW 11.460–61). See Hamilton 1998.

locke and british empiricism

295

in the mind’s existence (PHK 97–98). The concept of time has no legitimate application beyond the experience of succession from which it derives. (Hume invoked a similar view for different purposes at Treatise 35–37, 64–65.) No single episode could better illustrate Berkeley’s curious admixture of empiricism with Cartesian and Malebranchian metaphysics. Berkeley relies on empiricist analyses—of time, body, causation, and material substratum—on an “as-needed” basis, when it suits his metaphysical purposes. It is a mistake to detach Berkeley’s empiricism from its Cartesian and Malebranchian moorings.21 With this picture of Berkeley in place, I return to knowledge of the external world in Locke and Hume.

7. Epistemic Priority and Reification Some schematic stage-setting will facilitate the discussion. Distinguish between beliefs entirely about the content of one’s own, current sensory experience (for example, “it visually appears to me as if there is a table in front of me”) and beliefs about physical objects (“there is a table in front of me”). Descartes held that appearance statements are epistemologically basic: they can be justifiably believed independently of their inferential relations to any other beliefs. He also held that appearance statements are epistemologically prior to physical object statements: the justification for any belief in a physical object statement must ultimately derive from its inferential relations exclusively to beliefs in appearance statements. This is a constraint on material object statements globally, as a class. These commitments, embedded in Descartes’ First Meditation skepticism,22 generate foundationalist theories of knowledge of the physical world. Descartes also held that appearance statements are incorrigible, cannot be mistaken, and are certain (CSM 2.19: AT 7.29). These additional theses generate a classical or Cartesian foundationalist theory on which any justification for beliefs about physical objects ultimately derives from a bedrock of certainty. Locke agreed that appearance statements are basic and certain: “There can be nothing more certain, than that the Idea we receive from an external Object is in our Minds; this is intuitive Knowledge” (Essay IV.ii.14; cf. II.ii.1, II.xxix.5, IV.ix.3). For Berkeley, appearance statements are incorrigible: “[a man] is not mistaken with regard to the ideas he actually perceives” (DHP III, 238). So, too, Hume: “since all actions and sensations of the mind are known to us by consciousness, they must necessarily . . . be what they appear” (T 190; cf. 366). For the latter two figures, visual experience is also reified, held to consist in acquaintance with mental entities—Berkelian ideas, Humean perceptions— “sense data,” in the twentieth century. This ontological or metaphysical thesis,

21. Cf. Ayers 2005, 37, 50, 56. 22. See M. Williams 1986.

296

reflection and the stability of belief

prominent early in the figures’ main works (PHK 4; T 67), is superimposed on the doctrines of basicness and priority. Berkeley is explicit that beliefs based strictly on the senses are epistemically basic: “in truth the senses perceive nothing which they do not perceive immediately: for they make no inferences” (DHP I, 174–75; cf. III, 238). In such contexts, ideas are doubly “immediate”—a locus of noninferential certainty (PHK 18; DHP I, 204–5, II, 221, III, 238) and also objects of perceptual awareness (PHK 38, 56; DHP I, 195, III, 251, 262); a commitment to epistemic priority converges with reification. This is precisely Hume’s position. Reification and priority appear in a single sentence: “nothing is ever really present with the mind but its perceptions . . . , and . . . external objects become known to us only by those perceptions” (T 67; cf. 212). As with Berkeley, the noninferential certainty and epistemic priority (T 82) of appearance statements has come to be located in statements about perceptions “immediately present” (T 107, 119). Whether Descartes yoked reification to his foundationalism is controversial. Similarly, notwithstanding the foundationalist currents in Locke (Essay IV.iii.19; cf. IV.ii.1), his theory of perceptual experience is variously interpreted— as a reifying indirect realism but also as direct realism, as adverbialism, or as neutral among distinctively philosophical theories of perception.23 What accounts for the unambiguous reification in Berkeley and Hume? In the Treatise, reification is an unsupported presupposition. In the Enquiry, a scant paragraph (EHU 152) advances the nugatory “diminishing table” argument, which Thomas Reid demolished (EIP 178–84).24 “Causal gap” arguments for reification had been in the air.25 How could a visual experience at a perceiver’s location consist in direct awareness of a cause at a spatial and temporal remove? In Malebranche’s Search, the “immediate object” of perception is “the object closest to the mind, when it perceives something” (3.2.1.1/17). Reid attacked the argument as formulated by Samuel Clarke in 1716, and especially by William Porterfield, originally in 1739 (EIP 175–78).26 There are traces of these arguments in Locke’s observation that “external Objects be not united to our Minds, when they produce Ideas in it” (Essay II.viii.12); in Berkeley’s argument about the causation of sound (DHP I, 181); and in Hume’s passing comment that the senses cannot produce any “immediate intercourse between the mind and the object” (EHU 152). There is a more decisive route to reification in Berkeley. His adverbialist strains are far from thoroughgoing. Further, if realism is false, direct realism is false. When the existence of mind-independent material objects is denied 23. See Yolton 1984, chs. 1, 5; de Bary 2002, 105–29; Robinson 1988, ch. 1; and Chappell et al. 2004. 24. The attack, however, had to be renewed against sense data theories. See Pitcher 1971, 28–42. 25. Yolton 1984, chs. 3–4, and Lennon 1993, 251–73. 26. Reid’s objections were idiosyncratic. Russell’s revival of the argument, emphasizing the temporal gap (1912, ch. 3, and 1927, 197), led to renewed criticism (Armstrong 1961, ch. 11, and Pitcher 1971, 43–59).

locke and british empiricism

297

out of the gate, the objects of perceptual experience must be something else. The thought that perceptual objects reside in the Divine mind is a Malebranchian excess. Berkeley’s reification seems inevitable. Reification in Hume was to some extent vestigial. At the same time, whereas Berkeley emphatically rejects realism, in Hume, as we shall find in §11, realism gains little traction. It is thus convenient for him to treat objects of perceptual awareness as something other than extended surfaces. Epistemic priority, though not reification, was a shared doctrine among Descartes, Locke, Berkeley, and Hume. (There are refinements in §10 with respect to Locke.) The figures differ in their accounts of the types of inferential relationships to appearance statements that can serve to justify statements about physical objects. Descartes seeks to validate sense-perception via a proof of the existence of a nondeceiving God, appealing to clear and distinct perception, reason, to establish informative or “instructive” metaphysical propositions—for example, that there must be at least as much perfection in the cause as in its effect. Descartes relies on synthetic truths that are necessary and known a priori—not an option for Hume, and by and large not for Berkeley. Two inferential strategies remain. Berkelian phenomenalism seeks to rely on statements that are true by virtue of meaning, analytic, and hence a priori—to guarantee the existence of the physical world via a semantic, rather than a metaphysical, theory. The other empiricist strategy is to rely on inductive inference.

8. The External World and Inference to the Best Explanation Treatise I.iii, where Hume seeks to advance beyond Locke’s restrictions on sensitive knowledge, presupposes that the senses and memory supply knowledge, however fragmentary. How do we know objects exist at times they are perceived? At Essay IV.xi.4–7, Locke inquires into how visual and tactile experiences that are unavoidable, mutually confirming, and constant and regular are “produced” (Essay IV.xi.4–6). It is natural to construe Locke’s strategy in these passages in terms of theoretical induction, inference to the hypothesis that best explains the features in question. Thus Berkeley: “perhaps it may be thought easier to conceive and explain the manner of [our sensations’] production, by supposing external bodies . . . rather than otherwise; and so it might be at least probable there are such things as bodies” (PHK 19, emphasis added). Ironically, Locke stopped short of inferring the existence of material objects. He takes care to conclude only that sensory experiences have “exteriour” (Essay IV.xi.4–5) or “external” (IV.xi.6) causes, causes “without” (IV.xi.5, 7) the mind. (Reliabilist interpretations have difficulty explaining this restriction.) Belief in matter may be probable but does not rise to the threshold of certainty required for sensitive knowledge. Why the reticence to extend the argument to belief in matter? Berkeley raised an ad hominem objection: “the materialists . . . by their

298

reflection and the stability of belief

own confession are . . . unable to comprehend in what manner body can act upon spirit” (PHK 19). Although Locke did find body-to-mind interaction inconceivable (Essay IV.iii.6, 13–14, 28), he was likely more troubled by explanatory opacities internal to the best available physical theory, the “corpuscularian Hypothesis” (IV.iii.16). Cohesion of solid parts and the communication of motion are themselves difficult to conceive or comprehend (II.xxiii.23–29). This is one factor restraining Locke’s conclusion. Hume would have had no hesitation on this ground. He rejected the search for intelligibility, in the form of necessary connections, as in principle misguided. It is “sufficient satisfaction” (EHU 43) “to reduce” phenomena to some “ultimate springs and principles . . . [e]lasticity, gravity, cohesion of parts, communication of motion by impulse,” though beyond “any particular explication” (30). Hume’s objection to inference to the best explanation is more fundamental. Custom or conditioning explains the psychological irresistibility of beliefs arising specifically from enumerative induction, and hence (as a first approximation) their favored epistemic status. Beliefs resulting from inference to the best explanation are not “just,” as they are not conditioned and irresistible. This posture results from Hume’s attempt to integrate the psychology and epistemology of induction. Though no skeptic about enumerative induction, Hume, in Newtonian fashion, considers theoretical induction suspect. Hume applies this body of thought to explanationism: “could not [perceptions] arise either from the energy of the mind itself, or from the suggestion of some invisible and unknown spirit, or from some other cause still more unknown to us?” (EHU 152–53; cf. T 84). Inference to the best explanation requires opting for one of a myriad of hypotheses, independently of the psychological constraints imposed by repeated observation of similar pairs of conjoined objects. Further, though Hume is willing to employ Berkeley’s ad hominem objection (EHU 152–53), “any thing may . . . produce any thing” (164); a prioristic restrictions on causal relations, in the manner of Berkeley, do not narrow down the field. Locke’s development of his argument obscures the central difficulty. He takes the salient skeptical target to be a Cartesian dreaming hypothesis (Essay IV.ii.14, IV.xi.8), indeed that the causes of sensory experience are “Fancies” (IV.xi.5–6) resulting from “the Sport and Play of my own Imagination” (IV.xi.7). Locke thus pits unregulated mental activity against external causes that operate “regularly” (IV.xi.7). Even waiving the hypothesis of an unavoidable and regular Humean “energy of the mind” or dream, an “external” or “exteriour” cause need not be extended; it could be an “invisible . . . spirit”—Descartes’ powerful deceiver or Berkeley’s God—causing sensory experiences, in the absence of matter, in a constant or systematic way.27 Locke seems unperturbed; the certainty of external existents that are 27. Twentieth-century attempts to exploit inference to the best explanation in order to justify physical object statements, though elaborated in great detail, are susceptible to the same objection. See Russell 1912, ch. 2, and 1927, ch. 20; Broad 1914, ch. 4, and 1923, ch. 8; Price 1932, ch. 4; Reichenbach 1938, §14; and Ewing et al. 1945.

locke and british empiricism

299

“convenient or inconvenient,” that cause “Pleasure or Pain; i.e., Happiness or Misery” is “as great . . . as our Condition needs” (IV.xi.8; cf. IV.ii.14). Knowledge of particular facts about the material world is at some remove from an epistemology in the service of morality and religion.

9. Enumerative Induction and a Revived Explanationism For Hume, inference to the best explanation is a nonstarter. Enumerative induction can seem powerless here, too: “as no beings are ever present to the mind but perceptions; . . . we . . . can never observe [a conjunction] between perceptions and objects” (T 212); “experience is, and must be entirely silent” (EHU 153) on the question of external objects. Although Hume’s formulations rely on the metaphysical claim that perceptions are the only objects present to the mind, and hence rely on reification, his commitment to epistemological priority suffices to deliver the same epistemological result.28 To see this, consider the following enumerative argument: “whenever I have been appeared to as if there is a table in front of me, there has been a table in front of me; I am appeared to now as if there is a table in front of me; therefore, there is a table in front of me.” The argument does not assume a reifying account of appearance statements. Even so, the argument violates the foundationalist’s own constraint that the justification for any belief in a physical object statement (and hence in the conclusion) must ultimately be justified exclusively on the basis of appearance statements; the initial premise includes an undischarged statement about material objects. Within Hume’s ontology, “we may observe a conjunction or a relation of cause and effect between different perceptions” (T 212). Hume finds it conceivable, and hence possible, that perceptions exist when not perceived, separate from any mind (207; cf. 233). Suppose some perceptions do exist independently of being perceived. If so, they would possess a signal characteristic of bodies. Hume indeed stipulates that a “body” just is an object with a continued and distinct or independent existence (187–88; cf. EHU 151–52). Anticipating neutral monism, perceptions are not intrinsically mental. If causal inference can proceed from observed to unobserved perceptions, the existence of “body” is secured! The trick is to apply enumerative induction entirely within the realm of perceptions and to offer a reforming definition of “body.” Hume’s exploration of neutral monism was more than two centuries ahead of its time.29 Although none of Hume’s predecessors allowed unsensed perceptions, Locke inadvertently paved the way. He disparaged substrata and promoted the hypothesis of thinking matter (Essay IV.iii.6; Works

28. Cf. Jackson 1977, 147–51. 29. Neutral monism is found in Mach 1886; James 1904a and 1904b; and Russell 1921.

300

reflection and the stability of belief

4.31–36) and hence conscious states separate from immaterial souls. Berkeley thus pauses to fend off the worry that in his system the mind is “only a system of floating ideas” (DHP III, 233; cf. PN 579–81), a system of ideas separate from immaterial substance. This is a step en route to perceptions separate from mind altogether. Hume changes tack, deciding that “a very little reflection” (T 210) shows that perceptions do not, in fact, continue to exist unperceived. Hume nevertheless pursued neutral monism far enough—in his discussion of the “coherence” of perceptions—to discover a significant hurdle. Suppose Hume sees a four-log fire (4L), dozes off, and later observes a two-log fire (2L). Hume takes these observed objects, various stages of the “fire,” to be perceptions. He wants to rely on enumerative induction to infer the existence of other perceptions, the (unobserved) three-log fire (3L)—either as the unobserved effect of 4L or as the unobserved cause of 2L. But four-log fires typically induce him to doze; suppose he has observed 3Ls to follow 4Ls, or 3Ls to precede 2Ls, only 30 percent of the time. Yet, we think it overwhelmingly probable, not 30 percent likely, that an unobserved 3L exists. The observed statistical correlations do not lead to a sufficiently strong custom or habit to support this belief (197–98). Hume appeals to a mental “galley,” a principle of mental inertia that extends or enhances observed statistical regularities. Why would the mind do that? Hume’s most promising suggestion is that only the supposition of a “greater regularity among objects” (T 198) than we observe—the existence of an unperceived 3L—allows us to “account for,” to explain, such phenomena as 2L. Our hypothesizing nonstatistical regularities is a kind of inference to the best explanation after all—an inference to unobserved perceptions, in the interest of subsuming observations under explanatory regularities.30 This account faces a severe difficulty. In suggesting that the explanations under the enhanced regularities must be “conformable to my experience in other instances” (T 196), Hume presupposes that there is a straightforward way to decide, from within the confines of appearance statements, how to construct and apply explanatory generalizations exhibiting greater regularity. We had better not say that 2Ls are always preceded by (observed or unobserved) 3Ls—not so in cases where the porter reignited a poorly laid fire that burned out while Hume slept, or in cases where the porter laid only a two-log fire. We cannot locate robust regularities within the realm of appearance statements.31 This result poses a significant difficulty for positions that postulate an unobserved cause—whether Hume’s unobserved perceptions, extended bodies, Locke’s exterior objects, or Berkeley’s God32—to explain putative regularities in experience. It poses much the same problem for semantic

30. Cf. Bennett 2001, 2.294–95. 31. Cf. M. Williams 1977, 137–44. 32. Cf. Ayers 1984, 321–22.

locke and british empiricism

301

phenomenalism: “If someone were appropriately situated, they would be having table-ish sensory experiences”—but not if, for example, the person’s visual cortex is damaged, or his eyes are closed—conditions which must themselves be equivalent to counterfactuals supported by lawful regularities formulable in purely experiential terms.33 Phenomenalism never overcame these difficulties.34 These are reasons to think that if we are to secure perceptual knowledge of the physical world, the epistemic priority of appearance statements to physical object statements must be rejected. As we shall see shortly, the Enquiry can read as if Hume absorbed this moral. Such lessons can be slow to take hold. Mill, more than a century later, reverts to an attempt to ground belief in the external world—groups of permanent possibilities of sensation related by laws of succession—via associationistically generated, enumerative extrapolation from order within sensations (CW 9, ch. 11, and appendix to chs. 11–12).

10. Nativism and Basic Beliefs in External Objects We begin with a strand in Locke’s polemic against innatism. Locke contends that we are endowed with faculties sufficient for knowledge bearing on religious, moral, and prudential concerns; for this reason, Divine goodness does not require that God also implant innate beliefs (Essay I.iv.12; cf. I.i.5). Our prudential concerns include survival. Hume observes that “upon . . . removal” (T 225) of the faculty of causal inference, “human nature must immediately perish” (225). The operation of custom must therefore be instinctive (179), “permanent, irresistable, and universal” (225). A Lockean, admitting innate cognitive capacities, can absorb this claim; he can accept custom as an inborn faculty, adding that it operates instinctively. In the Enquiry, Hume elaborates the case for the role of instinct. The alternative is that employment of a faculty generating beliefs “necessary to . . . subsistence” (EHU 55; cf. 54) depends on reflection and argument. Two considerations, at least, tell against this: “reason . . . appears not, in any degree, during the first years of infancy” (55; cf. T 193) and “is slow in its operations” (EHU 55; cf. 39). For these reasons, “It is more conformable to the ordinary wisdom of nature to secure so necessary an act of the mind, by some instinct” (55; cf. 106, 108). In these passages, Hume has causal inference in view. In Enquiry XII, Hume extends this cluster of arguments to belief in “external” objects, an “external universe, which depends not on our perception” (EHU 151). This is the Treatise belief in “body”—“continued and distinct” entails “external” existence (T 188; cf. EHU 152). The belief in external objects arises “almost before the use of reason”; “even the animal

33. Cf. Ayer 1936 and 1940, and C. I. Lewis 1946. 34. See Chisholm 1948 and Sellars 1963.

302

reflection and the stability of belief

creation . . . preserve[s] this belief of external objects, in all their thoughts, designs, and actions” (EHU 151). These are broadly empirical arguments; the belief in external objects must result from instinct, because it is essential to the maintenance of unreflective creatures, such as animals and young children. This conclusion, taken on its own, leaves open the possibility that the belief in external objects is due to instinct simply because the belief is the product of the faculty of causal inference, operating instinctively. Hume rejects this option: the belief in external objects arises “without any reasoning” (EHU 151), meaning without theoretical or enumerative induction, without causal inference of any kind. The belief is the upshot of a primitive, that is, unexplained “natural instinct or prepossession,” a “blind and powerful instinct of nature” (151). This represents a major turnabout from the Treatise. Hume’s program in his treatment of coherence was to explain the psychological strength of the belief in “body” in terms of that of custom or causal inference. In the Enquiry, Hume abandons this project. In light of our discussion in the preceding section, we can speculate that he came to appreciate the poverty of regularities within experience. He retains the causal theory of knowledge, but in a context where beliefs in external objects can serve as starting points for causal inference. Hume thereby expands the range of beliefs that are epistemically basic. Since basic beliefs in external objects are necessary in order to survive, they must result from instinct. Since they do not result from a domain-neutral faculty of causal inference, applicable to objects of whatever kind, they must be the upshot of a specialized instinct to “repose faith in [the] senses” (EHU 151). This seems to commit Hume to innate belief, in the form of an innate disposition, triggered by perceptual experience, to believe in external objects. In Enquiry XII, the instinctive belief in body, however emaciated, is specifically about external objects. Locke is quite mistaken about the superfluity of innate belief.35 Reid pressed the case for instinct much more vigorously—though he was seemingly oblivious to his kinship with Hume.36 (Many of Reid’s arguments anticipate the resurgence in innatism since the 1960s.) Reid is emphatic that “[t]he belief” in, and hence “the very conception” of, “something external, extended . . . are equally parts of our constitution” (IHM 72; cf. 62–65 and EIP 210, 226–27). Though he inveighed against classical foundationalism, he agreed that appearance statements are certain (EIP 470; cf. 66, 233). His objection is ultimately to Descartes’ “requiring proof by argument of every thing except the existence of [his] own thoughts” (518), and of the material world in particular (519, 525)—to epistemic priority.37 Belief in external, even extended, objects—though not certain—is instinctive and epistemically basic (476–78). 35. Cf. Mackie 1976, ch. 7, and Carruthers 1990. 36. See Skorupski 1993, 11–14, and my 2007 [this volume, ch. 12]. 37. Wolterstorff 1987.

locke and british empiricism

303

Locke, on one interpretation, held a view in this vicinity. Michael Ayers maintains that the explanationist arguments at Essay IV.xi.4–7—billed by Locke as “concurrent reasons” (IV.xi.3)—merely supplement an independent justification for sensitive knowledge already in place. On Ayers’ interpretation, natural beliefs about external objects, albeit confined to external bare powers, carry independent, noninferential authority.38 (If, as Ayers sometimes suggests, an externalist semantic theory secures this result, it is difficult to see why Locke does not allow that beliefs about external objects are fully certain.) Basic beliefs in the perceptual sphere are not confined to appearance statements. Does expansion of the scope of basic beliefs in Hume’s Enquiry and in Reid, and perhaps in Locke, offer the prospect of securing our perceptual knowledge of the material world? Suppose its appearing to Hume as if a four-log fire exists does give rise to the basic belief that the logs, presumably only the observed portion of the logs, exist at that time. If so, the problem in §9 confronting Humean enumerativism reemerges: perceptual experience at best generates de facto, statistical regularities. This is the case whether the basic beliefs are about reified perceptions, as in the Treatise, or about temporal and spatial fragments of material objects. In either case, the available basic beliefs and accompanying correlations are too skimpy to support even mundane beliefs about the material world. This is a problem for Locke’s IV.xi.9 confidence, noted in §3, that belief in the existence of “Millions of Men” at times they are not observed is “highly probable” in virtue of its “conformity . . . with our own Knowledge, Observation, and Experience” (Essay IV.xvi.4). Even supposing perceptual knowledge of the existence of extended objects at times they are perceived, there is nothing close to “constant Observation” (IV.xvi.6) to support the belief—in our Humean example—in the unobserved, three-log fire. Hume perhaps sensed the difficulty; the instinctive belief in an external object carries the supposition that it “preserves its existence uniform and entire” (EHU 152). This is a highly schematic suggestion. Are the logs whole or rotten? For how long do they preserve their existence? Even an innate bias in favor of beliefs in solid, persisting objects does not generate a sufficiently rich stock of basic beliefs to answer such everyday questions. To do so, we require rough-and-ready background beliefs—not only about the particular circumstances (mostly unobserved) in which we find ourselves but also about the general character of the laws governing the world.39 These beliefs seem radically underdetermined by experience, even supplemented by instinctive belief in extended objects. Everyday beliefs about full-bodied objects depend for their justification on a background theory; they are not epistemically basic. These considerations exert pressure in the direction of holistic versions of empiricism—coherentist theories that, contrary to the foundationalist picture, deny basic beliefs altogether.40 38. Ayers 1991, 153–92, 302–3, and 1994. 39. Cf. Bennett 2001, 2.291–92, and Pears 1990, 168–82. 40. As in Neurath 1932 and Quine 1951.

304

reflection and the stability of belief

11. Berkeley’s Influence on Hume We have rushed forward two hundred years, without taking into account Berkeley’s influence on Hume. In Enquiry XII, Hume grafts an instinctive, constitutional belief in external objects onto the experience of what are in fact perceptions (EHU 152). In principle, he could have treated belief in extended objects in analogous fashion. Then why does Hume stop short at Lockean belief in mere externality? When Locke relies on inference to the best explanation, he has the standard for knowledge ratcheted up to certainty; he can be more certain of external, than extended, causes. Hume’s concern is different: “reason . . . [b]ereaves matter of all its intelligible qualities, both primary and secondary, . . . in a manner annihilating it,” “leav[ing] only a certain unknown, inexplicable something” (EHU 155). Removing primary qualities from our concept of external objects evacuates the belief of all respectable content. Berkeley provides the backdrop: perceptual relativity shows that bodies do not possess properties resembling experiences of secondary qualities (DHP I, 183–86); thus, colored expanses—redness, whiteness—exist only in the mind. Hume agrees (T 226–27). (The diminishing table argument relies on relativity for a different purpose, to argue against direct realism.) From this starting point, there are two Berkelian arguments for stripping external objects of primary qualities, and hence two routes against realism—if an object lacks a determinate shape and size, it is not so much as extended. This implication was of no concern to Berkeley, the immaterialist who believed the universe a community of spirits. For a Cartesian or Lockean realist, any argument to this conclusion would be a disaster. Berkeley contends that, given the variation in an object’s apparent color, the supposition that it possesses one determinate resembling color rather than another is arbitrary (DHP I, 185–86). One of the two antirealist paths generalizes this pattern of argument to primary qualities (cf. 187–92; cf. PHK 15). There is a strategy for blocking the generalized argument: it is not arbitrary to suppose that a billiard ball, for example, is spherical rather than cubical—not if we seek an economical physical theory to explain its causal interactions. Berkeley’s equanimity in suggesting that “the same arguments” apply to primary qualities ignores the Lockean, scientific realist argument for a primarysecondary quality distinction—that size and shape, but not properties resembling experiences of color, play an explanatory role in our best scientific theories.41 (Perhaps Berkeley had his eye more on Malebranche and Bayle.42) Unsurprisingly, Berkeley applies the perceptual relativity argument first to color, where it is on its strongest ground, and only later to shape. When Russell reproduced the argument, he followed the same tactical sequence.43 41. For this reading of Locke, see Mandelbaum 1964, 18–30; Alexander 1974; and Mackie 1976, 17–23. M. Wilson 1982 calls into question whether Berkeley ignored this reading. 42. See Popkin 1951a and McCracken 1983, 217–22. 43. Russell 1912, ch. 1.

locke and british empiricism

305

Hume sided with Locke. He allows that “the real table . . . exists independent of us” (EHU 152); “the various aspects of objects, according to their different distances . . . are only sufficient to prove . . . that we must correct their evidence by reason” (151). We “correct” the appearances “by considerations, derived from the nature of the medium, the distance of the object, and the disposition of the organ” (151; cf. Essay II.viii.21)—“general rules” governing the variation (cf. T 603, 632/1.3.10.12; EPM 227–28). Hume must hold that the best natural philosophy of his time motivates these corrections. Berkeley was an unabashed, unconflicted idealist; Hume was much more receptive to the (scientific) realist worldview. Yet, Hume signs on to the second Berkelian argument: we cannot conceive of primary qualities existing separately from, abstracted from, colored expanses together with other secondary qualities; therefore, primary qualities cannot possess such separate existence—they exist only in the mind (PHK 10; DHP I, 192–94). This is the argument that bereaves matter of primary qualities (EHU 154–55). It depends on assimilating conceivability to imageability.44 Whatever the extent of Locke’s commitment to imagism, he did not agree that impossibility follows from inconceivability (Essay IV.x.19).45 For Berkeley, the argument substantiates his claim that the doctrine of abstract ideas lies at the root of materialism (PHK 5). Why does Hume elaborate the argument at great length in the Treatise (T 227–31) and retain it in the Enquiry? Hume professed to find Berkeley’s views on abstraction “one of the greatest discoveries that has been made of late years in the republic of letters” (17)—effusive praise. And an opportunity for a three- to four-page display, early in the Treatise, of the fruitfulness of associationism—in explaining how a determinate image manages indifferently to represent disparate objects (20–24). This gives Hume a stake in applications of Berkeley’s views on abstraction. There is a second factor. Acknowledging that the imageability argument “is drawn from Dr. Berkeley,” Hume observes “that all his arguments, . . . admit of no answer and produce no conviction. Their only effect is to cause . . . momentary amazement and irresolution and confusion” (EHU 155n.). The Berkelian arguments against matter cannot result in sustained belief. “Nature is always too strong for principle” (160); the instinctive belief-forming mechanisms overwhelm “merely skeptical” (155n.) arguments. Hume was not content, as were many of Berkeley’s contemporaries, merely to dismiss his views.46 Hume exploits the imageability argument as a foil for his naturalistic rejoinder to Pyrrhonian arguments. This is not a development of Berkeley, but a substantive response to him.

44. But see Davidson and Hornstein 1984. 45. For discussion of Locke’s commitment to imagism, see Ayers 1991, 44–51; Chappell 1994; and Bennett 2001, 2.13–15. For Locke’s refusal to infer impossibility from inconceivability, see Jacovides 2002. 46. For attitudes toward Berkeley among his contemporaries, see Bracken 1958.

306

reflection and the stability of belief

Hume owes us a longer story. Reflection overturns both realism (whether direct or indirect) and neutral monism (§9)—every account of “body” or external objects Hume considers. Hume’s response, in part, is an elaboration of his psychology of mental conflict. The contradictions between reflection and instinct rein in our confidence in the refined arguments (EHU 161– 62)—such as the imageability argument—that contribute to them.47 And it is appropriate that nature should fashion instinct psychologically to trump reflection: “reason . . . at best is, in every age and period of human life, extremely liable to error and mistake” (55; cf. 106); instinct, though fallible (159), is a “mechanical tendency” or “mechanical power” (55, 108), less likely than reflection to go radically astray. Hume does think the ontological underpinnings of natural philosophy deeply problematic (cf. T 233, 366, 633). At the same time, he ultimately has less confidence in philosophical arguments for this result than in a cautious and modest investigation firmly grounded in instinctive beliefs and propensities (cf. EHU 161–62). Reflection is epistemologically inferior to instinct! Hume bases this provocative conclusion on his observation of mechanisms of belief formation, their interactions and natural functions, and their reliability within various domains. This empirical methodology anticipates many of the diverse naturalistic trends in epistemology over the last forty years.48

47. Passmore 1952/1968, ch. 7, and M. Williams 2004. 48. I am indebted to Victor Caston for invaluable suggestions, and to Lawrence Sklar for wide-ranging discussion. Special thanks to David Fate Norton, especially for exchanges in regard to Hume’s access to Locke’s Essay; to Jonathan Bennett, Lisa Downing, Daniel Kaufman, and Jennifer Nagel, for input on Locke’s philosophy and the Locke literature; to Kenneth Winkler, for the same on Berkeley; to George Pappas, for access to forthcoming work; and to Martha Bolton, Lewis Powell, Stephanie Rocknak, Matthew Stuart, and other participants at the fourth biennial Margaret Dauler Wilson Conference, for helpful questions. In his capacity as editor of Blackwell’s forthcoming A Companion to Locke, Matt Stuart also provided detailed and instructive comments.

locke and british empiricism

307

12 The Naturalisms of Hume and Reid

1. Some Puzzles consider the following picture in the rough. Some beliefs arise from our very constitution, from instinct; these beliefs, often irresistible, are justified even in the absence of arguments or reasons in their behalf. I think it fair to say that Thomas Reid, on the received view of his philosophy, held a position of this sort; when Reid defended the legitimacy of the beliefs of “common sense,” he had in view beliefs resulting from unavoidable and universal instinctive mechanisms, the faculties associated with his “first principles.” I think it also fair to say that David Hume, on what is today the received interpretation of his philosophy, held a position of this sort.1 In his pair of 1905 articles, “The Naturalism of Hume,” Norman Kemp Smith introduced the term “natural beliefs” as a label for irresistible, instinctive beliefs, arising from “the particular fabric and constitution of the human species” in Hume’s system.2 Kemp Smith consolidated his account in his 1941 book; his naturalistic interpretation has been the dominant force in the Hume literature ever since. But wait. Reid was one of Hume’s severest critics, and advanced his theory of commonsense beliefs as a constructive response to Hume’s skepticism. If the received interpretations of Hume and Reid, respectively, are so 1. [In this chapter, quotations of the Treatise and Enquiries are based on the volumes edited by Norton and Norton (NN) and Beauchamp (B-98, B-00) in the Clarendon edition of Hume’s works.] 2. For “natural belief,” see Kemp Smith, 1905, 151, 155, 156, 166, 171, 339. For “the particular fabric and constitution,” see 1905, 343, and 1941, 65, 147, 199; cf. 19–20, 27–28, 454, 458. The natural beliefs are “sanctioned by nature” (1941, 459; cf. 486, 564).

308

much as on the right track, Hume himself had proposed a constructive alternative to skepticism, one that downgrades the role of argument, emphasizing in its place the irresistibility of instinctive beliefs that originate in human nature. Common sense in Reid and natural belief in Hume seem very much of a piece. If so, what are we to make of Reid’s own conception of his relationship to Hume? It might seem that Reid either was deeply confused in this regard, badly misread Hume, or both.3 Is it at all plausible that Reid could have misread Hume? The Treatise presents severe difficulties for any interpreter. A work of apparently conflicting strands—at once empiricist, skeptical, and naturalistic—there is the problem of seeing how they fit together.4 Add to this that Hume scatters his most suggestive epistemological remarks, which are often incidental to one or another associationist or psychological discussion. Beyond these factors, the interpretation Kemp Smith attacked, the Reid-Beattie interpretation, originated in Hume’s own time. Hume parades reductive metaphysical doctrines— about body, the mind, personal identity, and necessary connection. He denies the necessity of the principle that every new existent or modification of existence has a cause. Late eighteenth-century figures regarded these positions as “skeptical.” Hume seeks to demolish the argument from design and the possibility of testimonial evidence for miracles. Whereas twentieth-century positivist readers welcomed Hume’s attacks on metaphysics, a substantial self or soul, and arguments for God, Hume’s applications of his doctrines to religion were a call to arms for his contemporaries.5 What is more, Reid’s arsenal included a large-scale historical theory, of which Hume was but a part. In Reid’s figure-by-figure survey of “the way of ideas” and related views, the chapter on Hume is shorter than those on Arnauld and Leibniz, and far shorter than those on Malebranche, Descartes, Locke, and especially Berkeley, who receives the longest discussion by a good margin. Although he reads stretches of Hume with care, Reid was not engaged in sympathetic interpretation. His investment in the story of the Cartesian or common way of ideas could easily have led him to distort Hume’s views, if not by misconstruing particular passages, by reading Hume selectively.6 As a general rule, commentators on Reid have found it convenient 3. I owe these questions to Matthew Davidson. De Bary allows that Reid misread Hume in some relevant respects (2002, 9–12). 4. Mounce can thus describe Kemp Smith’s work as the result of a “redistribution of emphases” (1999, 5). Passmore was perhaps the first to highlight the conflicting strands (1952/1968). M. Williams nicely shows how they feed competing interpretive traditions (2004, §III). 5. Norton contends that Reid differs critically from Hume in his providentialism, and hence in his commitment to truth, to the reliability of our cognitive faculties (1979). This is a complex matter. It is unclear that an appeal to God, or even to reliability, plays an essential role in Reid’s epistemology. Cf. Alston 1985 and Yaffe 2000. And Hume himself has been interpreted as a reliabilist (Schmitt 1992 and 2004). 6. Of course, much as Reid and later the positivists understood Hume through the filters of their own philosophical views, Kemp Smith’s reading might have been colored by his idealism. I argue in my 2009a [this volume, ch. 9], however, that Kemp Smith’s idealist convictions led him to underestimate the extent of Hume’s naturalism.

the naturalisms of hume and reid

309

to adopt his conception of Hume’s philosophy. This is understandable, where the project is to consider Reid’s historical position through his own eyes. Yet, an investigation of how Reid could misinterpret Hume might yield additional insight.

2. Induction One entrée into Reid’s affinity with Hume is by way of their treatments of induction. Reid, unlike Hume, frequently uses “induction” and “inductive” and appeals to an “inductive principle” (IHM 198, 199, 200). Hume writes of inference or reasoning about causes and effects, and he tends to assimilate inductive inference to causal inference.7 This opened the door to one of Reid’s criticisms of Hume on causation, that night following day, though inductively supported, is not genuinely causal (EIP 87). There is another difference: Reid conceives regularities as a language of nature, where the conjoined objects are signs of one another (IHM 190, 196, 200). It remains that both figures discuss a principle of the uniformity of nature, variously described, with Reid at times adopting Hume’s first Enquiry formulation (EHU 37) in terms of the future resembling the past (IHM 196; cf. 41 and EIP 489). Both are sensitive to a distinction between accidental and nonaccidental regularities (T 4, 104–5, 146–50, 175; IHM 41, 199–200; EIP 350, 374, 561). Both investigate the epistemology of induction. I distinguish three prongs in Reid’s views on this topic. First, Reid takes it that the inductive principle, the belief in “the continuance of the present course of nature” (IHM 196), cannot be founded in argument or derived from antecedent reasoning (IHM 196, 198; EIP 490).8 This claim is basic Humean doctrine. In Section IV of the first Enquiry and in Treatise I.iii.6, Hume famously advances a “negative argument” (EHU 34), that inductive inference cannot be founded either on demonstrative or probable argument. Reid thus credited Hume with being the first “who put [the] question” of “the ground” of the belief in continuance (EIP 490; cf. IHM 197), and with showing that it “is not grounded upon any antecedent reasoning” (EIP 490), not “founded either upon knowledge or probability” (IHM 197; cf. 198).9 Second, Reid attributes inductive beliefs to human nature—“the principle [that the future will be like the past] . . . is made a part of our constitution” (EIP 7. See Price 1940, 29, and 1969, 176–79; Passmore 1952/1968, 29–34; and Pears 1990, 71–72. 8. This is a point about justifying induction globally. Reid allows that an argument for a benevolent God (IHM 196), or an inductive argument based on induction’s “track record” (cf. EIP 489), can enhance the standing of inductive inference. On this latter point, Hume would agree. Cf. Broughton, citing Treatise 105 (2008, 296). 9. In the Inquiry, the belief in continuance “cannot be founded either upon knowledge or probability” (IHM 197; cf. 198). Elsewhere, Reid writes as if Hume’s point were merely that belief in a principle of uniformity does not derive from intuition or demonstration: “[Hume] has shewn clearly and invincibly, that [this principle] is neither grounded upon reasoning, nor has that kind of intuitive evidence which mathematical axioms have. It is not a necessary truth” (EIP 490).

310

reflection and the stability of belief

489); “It is an instinctive prescience of the operations of nature” (IHM 198; cf. 196). This instinctive principle is “universal among mankind” (EIP 490; cf. 464–69, 499). Of course, we do not believe that fire will be followed by smoke, absent relevant experience. Particular inductive beliefs are “the result of experience and habit” (IHM 38; cf. 197, 201).10 Presumably, some applications of the inductive principle, as with other commonsense beliefs (cf. IHM 172 and EIP 42, 96, 99, 264, 514–15, 551), are irresistible.11 Similarly, “we are under a necessity of assenting to [first principles]” (IHM 71), which are “necessary for our subsistence and preservation” (EIP 375; cf. 226, 238–39, 501). These positive claims, as applied to induction, are also Humean doctrine. Inductive inference is due to custom or habit, an instinctive principle. Hume declares in the Treatise, with inference based on custom in view, that “reason is nothing but a wonderful and unintelligible instinct in our souls” (T 179). This passage and related texts lie at the core of the Kemp Smith interpretation.12 Further, in Treatise I.iii.13, the operation of custom “cannot be prevented by [reflection]” (147). In I.iv.4, causal inference founded in custom is Hume’s example of “principles which are permanent, irresistible, and universal,” “unavoidable to mankind” (225). These principles “are the foundation of all our thoughts and actions, so that upon their removal human nature must immediately perish and go to ruin”; they are, as with Reid, “necessary . . . in the conduct of life” (225; cf. EHU 108).13 The emphasis on such features as irresistibility, taken to be the effects of a universal, unavoidable instinct, the constitution of human nature, is to be found in Reid and Hume alike.14

10. For this reason, it is wrong to think of particular inductive beliefs as universal; not everyone will have the experiences that trigger a particular application of an instinctive principle (cf. Wolterstorff 2001, 223). 11. Reid writes “that as there are some first principles that yield conclusions of absolute certainty; so there are others that can only yield probable conclusions; and that the lowest degree of probability must be grounded on first principles as well as absolute certainty” (EIP 456). Reid would recognize various degrees of confidence in applications of the inductive principle, depending on the strength of the evidence (cf. EIP 557, 560–61). 12. Kemp Smith 1905, 157; cf. 165, and 1941, 64, 100, 199; cf. 123–24. 13. James Mackintosh reports: “As the present writer observed [to Thomas Brown] that Reid and Hume differed more in words than in opinions, he answered, ‘Yes, Reid bawled out, We must believe in an outward world; but added in a whisper, we can give no reason for our belief. Hume cries out, We can give no reason for such a notion; and whispers, I own we cannot get rid of it’” (PEP 346n.2). In truth, Reid says loud and clear that we can give no reason for the belief in body (e.g., EIP 148–52, 517), and Hume says loud and clear that we inevitably believe in body (T 187, 218). Passages relevant to induction may be found earlier in §2. 14. Reid’s rejection of Hume’s associationist explanation of custom or habit, and hence of inductive inference (IHM 199–200), does not undermine the similarities in positive teachings. Reid writes: “[Hume] has endeavored to account for [the belief that the future will be like the past] upon his own principles. It is not my business at present to examine the account he has given of this universal belief of mankind; because, whether his account of it be just or not, (and I think it is not), yet, as this belief is universal among mankind, and is not grounded upon any antecedent reasoning, but upon the constitution of the mind itself, it must be acknowledged to be a first principle, in the sense in which I use that word” (EIP 490). Reid recognizes that the Humean explanation of inductive belief can be detached from the claim that it is universal.

the naturalisms of hume and reid

311

It is worth pausing to consider whether Reid was aware of these common, constructive doctrines. It is surprisingly difficult to show that he was. Reid writes: “[W]e agree with the author of the Treatise of human nature in this, That our belief of the continuance of nature’s laws is not derived from reason. It is an instinctive prescience of the operations of nature” (IHM 198). I am unsure whether Reid intends the point that the belief is instinctive to fall within the scope of the agreement. Similarly, although Reid says that the belief in uniformity is universal, he stops short of saying that Hume took the belief to be universal.15 One might think that Reid could hardly have missed these points in Hume. The arrangement of the Treatise is relevant here. Hume’s explicit statements that causal inference is instinctive and irresistible appear in I.iii.16 (T 179) and I.iv.4 (225). Reid never cites these sections, which are far removed from the negative argument of I.iii.6.16 I do not find passages where Reid recognizes the irresistibility of inductive beliefs in Hume. Beginning with the discussion of custom and repetition in I.iii.8, Hume implicitly develops an associationist explanation of the irresistibility of those causal inferences that amount to proofs, because based on perfect habits. The sections on probability—I.iii.11–13—consolidate Hume’s position. (See, for example, 124, 130–31, 134–35, 142–43, 147, 153–54.) Reid does not refer to Hume’s views in this stretch of Part iii. Since Reid had no patience for Hume’s associationism, he might not have tracked this line of argument. (In §6, I revisit the question of Reid’s recognition of irresistibility in Hume.) I turn to the third element in Reid’s views on induction. When Reid writes that the belief in the continuance of nature is a “first principle, in the sense in which I use that word” (EIP 490), he is making the epistemic claim that inductive beliefs are evident and at least probable. We “distinguish evidence into different kinds” (229; cf. 481 and IHM 32) corresponding to the different first principles, though the evidence they provide is a matter of degree (cf. EIP 228, 456, 481). Inductive inference provides evidence but cannot be founded in argument; its justification must be traced to the fact that it results from a universal, unavoidable, instinctive principle. For Reid, the negative claim that the inductive principle is not founded in argument does not impugn the epistemic credentials of inductive belief. Does Reid diverge from Hume at this juncture? He does, if the skeptical interpretation of Hume on induction is correct. This interpretation locates skepticism in the negative argument of Treatise I.iii.6 and first Enquiry IV: Hume maintains that inductive inference is not justified because it cannot be supported by (a non-question-begging) argument. On the opposing Kemp Smith interpretation, custom is a source of approved, “natural” beliefs. Much Hume

15. See the quotation in the preceding note. 16. In making this and related claims that follow, I have not consulted A System of Logic Taught at Aberdeen, 1763, by Dr. T. Reid (Edinburgh University Library, shelf mark Dk. 3.2, 1–101), a transcription from student notes by John Campbell. These contain many criticisms of Hume. For discussion, see Michael and Michael 1987 and Michael 1999. The first of these papers also includes extended excerpts.

312

reflection and the stability of belief

scholarship since the 1940s has been given over to a struggle between skeptical and naturalistic interpretations, with Hume’s position on induction a main front. In my view, literature in the 1970s and 1980s established the nonskeptical reading as decisively as possible in the face of complex textual data.17 Consider the Treatise, beginning with the context for the negative argument in I.iii.6. As early as I.iii.2 and I.iii.4, Hume advances a causal theory of reasoning about the unobserved: “the mind in its reasonings from causes or effects carries its view beyond those objects, which it sees or remembers” (T 82; cf. 73, 74, and EHU 26, 146). In I.iii.6, Hume examines “the nature of [the] inference” (T 88; cf. EHU 32) from what we sense or remember to an unobserved cause or effect. Hume writes within the course of the negative argument itself: “cause and effect . . . ’tis the only [connection or relation of objects], on which we can found a just inference from one object to another” (T 89). Though Hume finds that the inference is not founded in argument, his commitment to causal inference survives unscathed.18 Early in I.iii.8, he attributes causal inference to custom (102–3; cf. EHU 43). Later in the section, a person who stops his journey at a riverbank “foresees the consequences of his proceeding forward; and his knowledge of these consequences is convey’d to him by past experience” (T 103, emphasis added; cf. 104, 148). In I.iii.9, the discussion of the two systems of beliefs or “realities” (108, Hume’s emphasis) constitutes an extended restatement of the causal theory of knowledge. Hume writes that “the first of these systems is the object of the memory and senses” (108). The second system of beliefs is “connected by custom, or if you will, by the relation of cause or effect” (108); “’Tis this latter principle, which peoples the world, and brings us acquainted with such existences, as by their removal in time and place, lie beyond the reach of the senses and memory” (108, emphasis added; cf. EHU 55). In addition, the skeptical problem of induction implies that all inductive inferences are on a par, equally without justification. Even so, in I.iii.11, some causal inferences, for example, inference to the belief that the sun will rise, constitute “proofs” (T 124). In I.iii.12, Hume provides an inventory of “degree[s] of evidence” (153; cf. 130–31 and EHU 110, 117) that includes proofs and also probability, good inductive arguments that fall short of proofs (T 130–31, 142; cf. EHU 56–59). In I.iii.13, Hume contrasts “unphilosophical probability”

17. See especially Beauchamp and Mappes 1975, Connon 1979 (a paper presented in 1976), Winters 1979, Beauchamp and Rosenberg 1981, Arnold 1983, and Broughton 1983. I have to ignore many of the complexities here. Unsurprisingly, some rearguard actions are afoot. Penelhum 1992 and Winkler 1999 are notable in this regard. There is, of course, literature that advances the skeptical interpretation out of inertia, with little attention to the nonskeptical alternative. For recent discussions of the case against the skeptical interpretation, see Noonan 1999, 110–31; my 2002, ch. 2; Millican 2002, §§11–12; my 2006 [this volume, ch. 10], and my 2008. 18. Some commentators suggest that Hume, unlike Reid, saw the negative argument about induction as showing that we lack “philosophical grounds” (Vernier 1976, 31) for everyday beliefs, as revealing an “epistemic defect” (Rysiew 2002, 444). (This can give rise to a picture in which Hume appeals to irresistibility simply to repair this defect. Cf. §6.) Wolterstorff takes more care in saying that, for Hume, inductive beliefs “lack a certain kind of positive epistemic status” (1987, 403).

the naturalisms of hume and reid

313

(T 143) with beliefs legitimately based on the senses, memory, and causal inference; within this section, some causal inferences are “just and conclusive” (144). In I.iii.15, Hume provides eight “rules by which to judge of causes and effects,” “all the logic I think proper to employ in my reasoning” (175). There is no accounting for this extensive textual data on the hypothesis that Hume is a skeptic about induction.19 Reid quotes the Treatise in some two dozen contexts in the Essays on the Intellectual Powers of Man—but never any section of Part iii later than I.iii.7. Reid does not discuss or allude to the foreknowledge in the riverbank passage in I.iii.8, the two systems of realities in I.iii.9, the admission of inductive proofs in I.iii.11, the degrees of inductive evidence in I.iii.12, the distinction between philosophical and unphilosophical probability in I.iii.13, or the rules to judge of causes and effects in I.iii.15. (Reid endorses Newtonian, methodological rules—without mentioning Hume’s rules—as “maxims of common sense” at IHM 12.)20 Nor does Reid take up a now famous passage in I.iv.4: “One who concludes somebody to be near him, when he hears an articulate voice in the dark, reasons justly and naturally; tho’ that conclusion be deriv’d from nothing but custom” (T 225). This is in the paragraph that cites causal inference as an example of the “permanent, irresistible, and universal” principles of the imagination. How could it be that Hume’s epistemic endorsements of causal inference withstand his negative claim that induction is not founded in argument? In the first Enquiry as well as the Treatise, Hume’s approval of causal inference is in place prior to the negative argument. In both works, Hume interposes the claim that causal inference is due to custom, to instinct, after the negative argument and before numerous expressions of his continuing commitment to the positive epistemic status of causal inference. The obvious interpretive hypothesis is that Hume’s view is akin to Reid’s: causal inference, albeit not founded on argument, is justified, with its justification tracing to features of custom’s operation or origin in human nature. Kemp Smith’s work identified the irresistible and instinctual character of custom as a salient candidate for playing this role in Hume’s epistemology.21 19. There is additional evidence in the first Enquiry. In Sections X and XI—“Of Miracles” and “Of a particular Providence and a future State”—Hume maintains that the belief in miracles, based on testimony, and the argument from design depend on illegitimate inferences. He thus contrasts these illegitimate forms of causal inference with those that are legitimate. In these sections, causal inference can be justified: “One, who, in our climate, should expect better weather in any week of June than in one of December, would reason justly, and conformably to experience” (EHU 110; cf. 113). There are numerous references to “rules of just reasoning” or to a “just reasoner” (EHU 136, 139, 142, 145). 20. Nor is there mention of Hume’s rules in the section “Of Newton’s Rules of Philosophizing” in Reid’s manuscript, “Materialism.” See PRLS 182–92. 21. A number of commentators have taken Hume’s and Reid’s epistemologies to be quite similar: Popkin, most emphatically in 1967, 7.456–57, but also in 2005, 9.56, and Craig 1990. In §4, I show that the similarity is even stronger than has been recognized; in §§5–7, I address the question of how Reid could have failed to appreciate his common ground with Hume. Mounce 1999, ch. 1, has taken some steps in the same direction.

314

reflection and the stability of belief

3. Memory It is instructive to consider Reid’s treatment of Hume on memory against this background. Reid is confident that the way of ideas leads to skepticism about the past: “For since ideas are things present, how can we, from our having a certain idea presently in our mind, conclude that an event really happened ten or twenty years ago corresponding to it?” (EIP 290; cf. 476). Reid attributes this worry to Hume: “It does not appear to have occurred either to Locke or to Berkeley, that their system has the same tendency to overturn the testimony of memory as the testimony of the senses. Mr. Hume saw further than both” (290). Reid writes in the Inquiry: “Zeno endeavoured to demonstrate the impossibility of motion; Hobbes, that there was no difference between right and wrong; and [Hume] that no credit is to be given to our senses, to our memory, or even to demonstration” (IHM 21). In another passage, Reid is more circumspect: “Mr. Hume has not, as far as I remember, directly called in question the testimony of memory; but he has laid down the premises by which its authority is overturned, leaving it to his reader to draw the conclusion” (EIP 475). Hume never does call memory into question, though Reid does not pause to ask why not.22 The question is pressing, given Reid’s contention that Hume proceeds with the “aim of establishing universal scepticism” (290; cf. IHM 20). Hume writes in Treatise I.iii.5: “it be a peculiar property of the memory to preserve the original order and position of its ideas” (T 85; cf. 9)—hardly skeptical in tone. This should be unsurprising. The causal theory of knowledge of the unobserved emerges in I.iii.2 and I.iii.4. Causal inference could not extend our knowledge beyond the observed unless the senses and memory are sources of knowledge. Hume’s causal theory of knowledge presupposes the positive epistemic standing of memory.23 Reid suggests that the way of ideas saddles Hume with the view that the evidence of memory must be supported by argument (EIP 289–90, 475–76). In fact, Hume, like Reid (455), gives an infinite regress argument for basic beliefs, and for noninferential memory knowledge in particular: When we infer effects from causes, we must establish the existence of these causes; which we have only two ways of doing, either by an immediate perception of our memory or senses, or by an inference from other causes . . . till we arrive at some object, which we see or remember. ’Tis impossible for us to carry on our inferences in infinitum; and the only thing, that can stop them, is an impression of the memory or senses, beyond which there is no room for doubt or inquiry. (T 83; cf. 97) 22. One passage might be thought to cut the other way. Hume seems concerned that memory is founded on enlivening, a “trivial” property (T 265). For Hume, any worry here is a general one about “[t]he memory, senses, and understanding.” See also note 36. 23. Goldman—I think unaware of the causal theory of knowledge in Hume—took the approach further, noting that memory and perception are themselves causal processes, and hence subsumable under the theory (1967).

the naturalisms of hume and reid

315

There is no room for doubt about the existence of the object we remember because memory, “recent and fresh” memory not yet subject to decay (143–44; cf. 85), is irresistible. Reid also restricts the highest degree of “certainty and evidence” to cases where memory is “recent . . . and fresh” (EIP 42). (In context, Reid is discussing memory of prior conscious states.) Hume writes that demonstration is “irresistible” early in the Treatise (T 31). In the I.iii.13 survey of degrees of evidence, “force and . . . vivacity are most conspicuous in the memory; and therefore our confidence in the veracity of that faculty is the greatest imaginable, and equals in many respects the assurance of a demonstration” (153). Causal inference constitutes “[t]he next degree of these qualities” (153) and is thus “irresistible” at I.iv.4. The irresistibility of memory is essential to Hume’s psychological explanation of the irresistibility of causal inference—this because the strength of a belief in an unobserved object is transmitted from the strength of the belief of the senses or memory from which it is inferred. When causal inference proceeds from a perfect habit—frequent observation of a constant conjunction— the strength of an initial belief in an observed object is transferred nearly in its entirety to the idea of the unobserved. The resulting belief can be irresistible only if the initial belief based on the senses or memory is irresistible. These doctrines are set out in associationist terms—where “vivacity” and “force” are identified with degree of confidence and assurance—in sections 5, 6, 8, and 12 of Part iii of Book I. (Reid of course rejects Hume’s associationist account of memory—at IHM 28–29, 197–98—but the claim that memory is irresistible can be detached from the associationism.) For Hume as well as for Reid, belief based on memory is often irresistible and is a source of evidence and knowledge, even absent supporting argument.

4. The Proper Object of Epistemology The similarity of Reid’s and Hume’s views about memory and induction are symptomatic of a deeper agreement about epistemology as an enterprise— less anachronistically, about one objective of a theory of the human mind. Reid’s criteria for discerning first principles include “the consent . . . of the learned and unlearned” (EIP 464; cf. 461, 499); beliefs about the unobserved held by both the “simple rustic” and the philosopher “are built on the very same ground” (561). It is also a mark of first principles that beliefs “appear so early in the minds of men, that they cannot be the effect of education, or of false reasoning” (467; cf. 483). Thus, “children and idiots have [the] belief [that the future will be like the past] as soon as they know that fire will burn them. It must therefore be the effect of instinct, not of reason” (IHM 196). So, too, with animals; inductive inference “is the effect of a principle of our nature, common to us with the brutes” (50; cf. 13). Reid collates these points: “The language of nature is the universal study; and the students are of different classes. Brutes, idiots, and children, employ themselves in this study” (200). Similarly, “Perception . . . implies no exercise of reason; and is common

316

reflection and the stability of belief

to men, children, idiots, and brutes” (173). With respect to first principles, “There is no searching for evidence, no weighing of arguments; the proposition is not deduced or inferred from another” (EIP 452; cf. 470, 476)—not if we are to account for the knowledge of children and animals. All these themes appear in Hume. In Treatise I.iv.2, Hume notes that “whatever convincing arguments philosophers may fancy they can produce to establish the belief of objects independent of the mind, ’tis obvious these arguments are known but to very few, and that ’tis not by them, that children, peasants, and the greatest part of mankind are induc’d” (T 193) to believe in body. He writes in I.iii.16, “Of the reason of animals”: Beasts . . . can never by any arguments form a general conclusion, that those objects, of which they have had no experience, resemble those of which they have. ’Tis therefore by means of custom alone, that experience operates upon them. All this was sufficiently evident with respect to man. But with respect to beasts there cannot be the least suspicion of mistake. (178) It was evident with respect to man, on two counts. First, in I.iii.6, belief in uniformity cannot be founded on argument. Second, in I.iii.6 and I.iii.8, “custom operates before we have time for reflection,” without “a moment’s delay,” “without reflecting on it” (104; cf. 93); “the imagination of itself supplies the place of . . . reflection” (93). Hume observes: When any hypothesis . . . is advanc’d to explain a mental operation, which is common to men and beasts, we must apply the same hypothesis to both. . . . The common defect of those systems, which philosophers have employ’d to account for the actions of the mind, is, that they suppose such a subtility and refinement of thought, as not only exceeds the capacity of mere animals, but even of children and the common people in our own species. (177) Hume, like Reid, insists on the continuity in human and animal understanding. This theme is prominent in three of the twelve sections of the first Enquiry, even though it is a much truncated work. (In the Treatise, Hume withholds discussion of the understanding of animals until I.iii.16.) Hume writes in the final paragraph of Section IV: “It is certain that the most ignorant and stupid peasants—nay infants, nay even brute beasts . . . expect a similar effect from a cause, which is similar in its sensible qualities and appearances”; “this conclusion” does not arise “by any process of argument” (EHU 39). In the final paragraph of Section V: “reason . . . appears not, in any degree, during the first years of infancy”; causal inference results from “an instinct” that “nature” has “implanted in us” (55). Section IX, “Of the Reason of Animals,” reiterates and consolidates these points. As in the Treatise, a theory that explains “the operations of the understanding” in humans “will acquire additional

the naturalisms of hume and reid

317

authority, if we find, that [it] . . . explain[s] the same phænomena in all other animals” (104). Specifically, “animals, as well as men, learn many things from experience, and infer, that the same events will always follow from the same causes” (105). Hume observes: It is impossible, that this inference of the animal can be founded on any process of argument or reasoning. . . . For if there be in reality any arguments of this nature, they surely lie too abstruse for the observation of such imperfect understandings. . . . Animals, therefore, are not guided in these inferences by reasoning: Neither are children: neither are the generality of mankind, in their ordinary actions and conclusions. (106) Then, in the final paragraph of the section: “experimental reasoning itself, which we possess in common with beasts, and on which the whole conduct of life depends, is nothing but a species of instinct” (108). Reid does not cite or allude to either of Hume’s sections on the reason of animals. Hume and Reid nonetheless agree that an explanation of the knowledge of reflective adults must also explain that of nonreflective adults, children, and brutes. A number of basic cognitive processes—memory and inductive inference among them—are common to humans and animals.24 This constraint exerts pressure in the direction of broadly externalist theories of knowledge. Reid and Hume do not merely maintain that we can have knowledge even though no argument is available; were argument or reflection required, they both insist, we could not explain the knowledge of humbler creatures. Theories of empirical knowledge that demand even the capacity to produce arguments, to elaborate reasons, are nonstarters.25 Classical foundationalist theories are ruled out of court.26 The most obvious

24. Both Hume (EHU 107n.) and Reid (IHM 200) allow that adult humans can do better cognitively than children or animals and offer explanations for this difference. Hume, I believe, would emphasize the availability to reflective adults of general rules, that is, second-order habits or customs. (See §7.) Reid notes that animals lack the power of abstracting and generalizing (EIP 388). 25. Similar considerations would rule out coherence theories—positive coherence theories that require for justification some nontrivial degree of coherence or interconnectedness among beliefs. I do not believe, however, that such theories were in play. A negative coherence theory, on which all beliefs are prima facie justified, is perhaps a nonexternalist option available to Hume. See my 2001b [this volume, ch. 5]. For interpretations on which Reid holds that commonsense beliefs, at least, possess intrinsic, though defeasible, justification, see Lehrer 1974, 101–2, 104; Alston 1985; and Yaffe 2000. 26. I think Wolterstorff correct that Reid is more fundamentally opposed to classical foundationalism than to the “representationalist” component of the way of ideas (1987, 399–400, 405–6) but mistaken that “Reid’s significance lies in the fact that he is the first great critic of . . . ‘classical foundationalism’” (406). Hume had targeted foundationalism, though Reid went about the critique more self-consciously and systematically. See also §5. De Bary also takes the view that Reid’s antifoundationalism is more fundamental than his opposition to the way of ideas (2002, 4, 18, 86). For Wolterstorff’s most recent views on Reid’s opposition to representationalism and foundationalism, see his 2001.

318

reflection and the stability of belief

alternative is the class of externalist theories in which knowledge is the result of the operation of belief-forming mechanisms that are suitably instinctual, or adaptive and properly functioning, or reliable, and so forth.27 Reid shows no recognition of this antifoundationalist alignment with Hume.

5. Epistemic Reduction and Psychological Explanation Let me turn to some differences between the naturalisms of Reid and Hume, of interest in themselves, that might have contributed to Reid’s blind spot. I have mentioned Hume’s reductionism (§1). Reid faults this tendency in Hume’s metaphysics, where “bodies . . . are nothing but ideas in the mind” and “what we call a mind is nothing but a train of ideas” (EIP 173; cf. 63, 162). Hume’s metaphysics is an offense to substance but also to dualism. In psychology, there is Hume’s reduction of mental processes to association. In epistemology, there is the reduction of knowledge, beyond the senses and memory, to causal inference: “By means of [the] relation [of cause and effect] alone we can go beyond the evidence of our memory and senses” (EHU 26; cf. T 73–74). Hume pursues the epistemic reduction in a variety of contexts. For example, “our faith [in human testimony] arises from the very same origin as our inferences from causes to effects, and from effects to causes” (T 113). For Reid, “a disposition . . . to believe what [others] tell us” is an “original principle” (IHM 194).28 Or consider other minds. In the Treatise, sympathy converts a belief that another person is experiencing a specific feeling into the feeling itself (T 316–20, 385–86). How do we acquire the initial belief? Hume writes that it is “the relation of cause and effect, by which we are convinc’d of the reality of the passion, with which we sympathize” (320; cf. 319, 576). Causal inferences about the experiences of others, based on observed conjunctions in one’s own case, are a precondition for the operation of sympathy. (This is another juncture where Hume is no skeptic about causal 27. For more on Hume’s externalism, see especially my 2006 and also 2008. For a proper function interpretation of Hume, see Craig 1987, 81, and Wolterstorff 1996, 166n.6; for Hume as a reliabilist, see Dauer 1980; Costa 1981; and Schmitt 1992, ch. 3, and 2004. For a proper function interpretation of Reid, see Wolterstorff 1987, 409–10, and 2001, 208–9, and Greco 2004, §3; for Reid as a reliabilist, see Alston 1985 and de Bary 2002; and as an externalist, see Van Cleve 1999. The convergence in the range of interpretations is a symptom of the externalist tendencies in both figures. 28. In the first Enquiry, Hume is more guarded: “This species of reasoning, perhaps, one may deny to be founded on the relation of cause and effect. I shall not dispute about a word. It will be sufficient to observe, that our assurance in any argument of this kind is derived from no other principle than our observation of the veracity of human testimony, and of the usual conformity of facts to the reports of witnesses” (EHU 111). The resulting position, which reduces testimonial evidence to inductive (rather than specifically causal) evidence, remains one with which Reid would disagree. Wolterstorff takes note of Reid’s opposition to Humean reductionism specifically in connection with testimony (2001, 164; cf. 64).

the naturalisms of hume and reid

319

inference.) For Reid, knowledge of the conscious states of others depends on “[a]nother first principle, . . . That certain features of the countenance, sounds of the voice, and gestures of the body, indicate certain thoughts and dispositions of mind” (EIP 484; cf. 486). Similarly, the burden of Section XI of the first Enquiry is that any cogent argument from the order in nature for the existence of God must be based on causal inference (EHU 135–36). For Reid, belief in a designing author of the universe relies neither on a general inductive principle nor on the first principles of contingent truths that apply to other minds. He invokes an additional, “metaphysical” (EIP 503) first principle, a “necessary truth” (507): “effects which have all the marks and tokens of design must proceed from a designing cause” (507; cf. 503). Though Reid grants that “we ought not to multiply [original principles] without necessity,” he maintains that the first principles “are more in number than is commonly thought” (EIP 349). In the case of inference to other minds, he seems in a rush to establish a distinct principle. Reid abruptly rules induction from one’s own case out of court: “But how shall experience instruct us when we see the sign only, when the thing signified is invisible? Now this is the case here; the thoughts and passions of the mind, as well as the mind itself, are invisible” (485–86; cf. LRF 184–85). He does not so much as consider the possibility that the fact that conscious states are visible in our own case might be sufficient to ground the inference inductively. Reid does not find his reliance on a good dozen first principles troubling (EIP 493).29 In Reid’s view, “the evidence of sense, the evidence of memory, and the evidence of the necessary relations of things, are all distinct and original kinds of evidence, equally grounded on our constitution; none of them depends upon, or can be resolved into, another” (IHM 32; cf. EIP 193). This generalizes: “I am not able to find any common nature to which [the different kinds of evidence] may all be reduced”—except that “they are all fitted by Nature to produce belief in the human mind” (EIP 229).30 This is a stipulative claim about what is to count as a first principle. Reid is open to the possibility that “we find some more general principle into which” a putative first principle “may be resolved” (IHM 61). (Here he is discussing a first principle of perception.) Descartes’ and Locke’s attempts to identify a common nature—clear and distinct perception, the perception of the agreement and disagreement of ideas—are nevertheless wrongheaded (EIP 229). Reid might have added to his list Hume’s reduction of knowledge, beyond the senses and memory, to causal inference, had he recognized this aspect of Hume’s program. 29. My sketchy inventory of Reid’s first principles does not include those about one’s own mind (principle 2), identity over time (principle 4), agency (principles 6 and 10), the metaprinciple, that natural faculties are not fallacious (principle 7)—see §6—and life and intelligence in humans (principle 8). 30. Greco has emphasized that Reid’s foundationalism is “‘broad’, in the sense that he allows a wide variety of sources of both foundational and non-foundational knowledge,” and that “none of these sources is reducible to the other” (2004, 148). Greco takes note of “[o]ne aspect of this independence,” that one faculty “does not need the vindication of” another (149). The nonreducibility I have in view is that one faculty is not a special case of another.

320

reflection and the stability of belief

First principles cannot be resolved into one another epistemically. Nor can they be explained psychologically.31 According to Reid, Hume is “far from conceiving [our belief in continuance] to be an original principle of the mind” (IHM 197). Reid’s ground for this claim is that Hume “endeavours to account for [continuance] from his favourite hypothesis” (197; cf. 199); “by [Hume’s] system, three laws of association, joined to a few original feelings, explain the whole mechanism of sense, imagination, memory, belief” (22). Reid is not merely objecting to associationism; he insists that “Sensation and memory . . . are simple, original, and perfectly distinct operations of the mind” (29). Since such basic mental operations as memory and belief are simple, they do not admit of definition or analysis (31; cf. 29, 167). Further, “no philosopher can give a shadow of reason” why sensation and memory should compel belief, “but that such is the nature of these operations” (28; cf. EIP 87). Here we encounter Reid’s theme, emphasized by Nicholas Wolterstorff, of the darkness or mystery of the mind.32 “In all our original faculties, the fabric and manner of operation is . . . beyond our comprehension” (EIP 394; cf. 193). In calling a principle “original,” Reid often means that it is not only innate but also “unaccountable” (IHM 41; cf. 15 and EIP 255, 349, 394).33 This perhaps got in the way of his acknowledging that custom in Hume functions as a first principle in virtue of being instinctive and innate.34 In claiming that psychological mechanisms are beyond our grasp, Reid might be calling attention to the gulf between human and Divine understanding or, as Wolterstorff suggests, to the need for practical trust in our 31. Reid’s posture toward unifying accounts, whether epistemic or psychological, of first principles is intertwined with his attitude toward inference to the best explanation. Reid allows that “To love simplicity, and to be pleased with it wherever we find it, is no imperfection, but the contrary. It is the result of good taste” (EIP 530). Yet, he will follow simplicity only so far: “There is, without doubt, in every work of Nature all the beautiful simplicity that is consistent with the end for which it was made. But if we hope to discover how Nature brings about its ends, merely from this principle, that it operates in the simplest and best way, we deceive ourselves, and forget that the wisdom of Nature is more above the wisdom of man, than man’s wisdom is above that of a child” (531). 32. Wolterstorff 2001, esp. 255–61. In the case of memory: “The knowledge which I have of things past by my memory, seems to me as unaccountable as an immediate knowledge would be of things to come. . . . I find in my mind a distinct conception and a firm belief of a series of past events; but how this is produced I know not. I call it memory, but this is only giving a name to it; it is not an account of its cause” (EIP 255). 33. Cf. DeRose 1989, 326. Wolterstorff cites the passage at IHM 19 approvingly, apparently agreeing that if a principle can be explained associationistically, it is not original (2001, 198). Earlier (and in a different context), Wolterstorff glosses “original” as “innate, indigenous” (63). The fact that a principle can be explained associationistically does not entail or imply that it is not innate, grounded in the constitution of human nature. 34. One sense of this claim emerges when Reid writes of finding “the simple and original principles of [man’s] constitution, of which no account can be given but the will of our Maker” (IHM 15). If a principle is unaccountable, we cannot explain why it obtains, apart from appeal to God’s will. Hume would agree that the inductive principle, for example, might be unaccountable in this sense. Hume is dubious that any “ultimate reason” can be offered for the instinctive operation of custom (T 179), ceding the explanation of its operation to those “who delight in the discovery and contemplation of final causes” (EHU 54–55).

the naturalisms of hume and reid

321

faculties. My concern is with the claim’s importance to Reid in supporting his proliferation of first principles. In Book I, Hume applies a test of reflective approval to the understanding. Consider the belief in a substantial self or soul. For Reid, this belief, arising from a first principle, is unaccountable (EIP 472–74). For Hume, the belief results from a multistage psychological process. We attribute strict identity to perceptions that, though variable and interrupted, are resembling; this is a “confusion and mistake” (T 254), rooted in a propensity to confound resembling ideas and resembling mental dispositions (253–54). When we notice the variation and interruption in the impressions, and thus recognize that they cannot be strictly the same, we suppose or feign an unobservable soul as the locus of the strict identity that we had been inclined to attribute to observable perceptions. The supposition of a soul is a “fiction” that functions to “disguise” the variation (254; cf. 255). Hume lays bare the operative psychological mechanism and regards it as debunking.35 Reid agrees that it would be disquieting, were Hume’s explanation correct. Reid, of course, has general objections to associationism: associationism cannot account for the different propositional attitudes (IHM 30; EIP 197–98, 401); Hume’s three laws of association are not sufficient (EIP 347– 53, 377); and so forth. Reid takes more specific issue with Hume’s account of belief in the existence of the mind as a fiction. He disparages the possibility that “those inferences which we draw from our sensations, namely, the existence of a mind, and of powers or faculties belonging to it, are . . . mere fictions of the mind, which a wise man should throw off as he does the belief of fairies” (IHM 37). In the opening chapter of the Inquiry into the Human Mind, Reid complains: [I]f [the mind] is indeed what [Hume] makes it, I find I have been only in an inchanted castle, imposed upon by spectres and apparitions. I blush inwardly to think how I have been deluded; I am ashamed of my frame. . . . O Nature, to put such tricks upon a silly creature, and then to take off the mask, and shew him how he hath been befooled. (22; cf. 35) Reid finds Hume’s account of the principles of the mind wanting; they fail to win reflective approval.36 Reid’s resistance, however hard and fast, to the psychological investigation of first principles forestalls subjecting his own account of the mind and its faculties to a similar test. To the extent that psychological mechanisms are opaque, they cannot be unmasked. 35. For discussion, see my 2001a. 36. One aspect of Reid’s thought is perhaps that vivacity is just a kind of spell. (Reid might think that any psychological account of the operation of first principles would strike us in this way.) Hume can seem to have this worry, at least in passing, when he observes in Treatise I.iv.7 that all belief depends on enlivening and vivacity, “which seemingly is so trivial” (T 265). Hume’s misgiving, however, does not arise in the course of his account of belief in I.iii.7–8, or elsewhere in Part iii; it is new to I.iv.7, where Hume tends to exaggerate and indulge broadly skeptical outcomes. For a different view, see Garrett 1997, 213.

322

reflection and the stability of belief

The test of reflective approval, implicit in Hume’s treatment of the belief in souls, is explicit in the conclusion of Book III of the Treatise. The mind should be “able to bear its own survey” (T 620), should be able to pass a test of reflexive self-scrutiny.37 The case at hand is the “sense of morals” (619). In the Treatise, the moral sense depends on the operation of the associationist mechanism of sympathy. Hume writes: “this sense [of morals] must certainly acquire new force, when reflecting on itself, it approves of those principles, from whence it is deriv’d, and finds nothing but what is great and good in its rise and origin” (619). Hume adds: “Those who resolve the sense of morals into original instincts of the human mind, may defend the cause of virtue with sufficient authority; but want the advantage, which those possess, who account for that sense by an extensive sympathy with mankind” (619). Hume has Hutcheson in view; if the moral sense is original, he is in no position to approve the principles from whence it is derived. Reid’s first principles cannot suffer the corresponding disadvantage. He forecloses any appeal to empirical psychology to call the epistemic status of any of his first principles into question. This is a breach in Reid’s naturalism.38 There is a second way in which the doctrine of the incomprehensibility of the mind shelters Reid’s multiplication of first principles. Reid individuates first principles in terms of their inputs and/or outputs—for example, perception of the countenance, voice, and gestures “in,” beliefs about other minds “out.” If the operative mechanisms are for us a black box, empirical psychology is powerless to support a reduction by resolving one first principle into another. Nor can we compare mechanisms in order to spot ones that might strike us as outliers. As it happens, Hume relaxes his reductionism in the Enquiries. The Treatise’s associationist explanation of sympathy gives way to humanity and fellow-feeling: “It is not probable, that these principles can be resolved into principles more simple and universal” (EPM 219–20n.). The Treatise’s multipage associationist explanation of probabilistic inference (T 127–30, 133–38) gives way to beliefs that arise “immediately, by an inexplicable contrivance of nature” (EHU 57). The Treatise’s fifteen-page explanation of why we believe in objects that have a continued and independent existence (T 194–210)—the discussion of coherence and constancy—also collapses into a primitive instinct: “It seems evident, that men are carried, by a natural instinct or prepossession, to repose faith in their senses; and that, without any reasoning, or even almost before the use of reason, we always suppose an external universe, which depends not on our perception” (EHU 151). This stunning turnaround sweeps away Hume’s attempt to ground the belief in body in causal inference. In explaining the belief in the Treatise, the only resources Hume allows himself are beliefs about impressions. He interprets the 37. For this strand in Hume, see Korsgaard 1989 and 1996, esp. 51–66, and Baier 1991. 38. Daniels, noting that Reid criticizes associationist accounts of the coordinated motion of the eyes (IHM 112–14), expresses some puzzlement as to why he does not offer an alternative explanation (1976, 36–37).

the naturalisms of hume and reid

323

Part iii formula that all knowledge is based on “the senses” (T 73, 108; cf. 74), memory, and causal inference to mean that all knowledge is based on sense impressions, memory, and causal inference (191, 193). The inputs to causal inference are beliefs, based on introspection or consciousness, about internal impressions, together with beliefs based on memory about past impressions. In cases of coherence—as in viewing a fire burning down—impressions “have a regular dependence on each other” (195); the belief in body “arises . . . from custom in an indirect and oblique manner” (197), and hence is due to “a kind of reasoning from causation” (195). Hume’s epistemic reductionism is at work in his treatment of coherence. In the first Enquiry, where there is no explanation of the belief in body, constancy and coherence alike drop out of the picture. The belief in body is an original, unaccountable instinct, coordinate with—in no way reducible to—custom or causal inference. This could be Reid (cf. EIP 476–77). Hume falls back, in Reid’s terminology, on first principles for consciousness, memory, causal inference, and perception of body—thus taking a step in the direction of Reid’s tolerance of multiple first principles. Hume’s uncompromising reductionism in the Treatise might have contributed to Reid’s failure to appreciate their common ground. Taking the inputs to causal inference to be beliefs about impressions, conscious states, has a classical foundationalist flavor. If one follows this interpretive lead and further observes that Hume has no arguments on offer to ground memory or causal inference, the skeptical reading is inevitable; Hume is a foundationalist who fails to secure knowledge on his own terms. As we have seen (§§2–4), however, Hume admits causal inference and memory as sources of evidence and knowledge on externalist grounds. The Treatise’s insistence on impressions as a starting point, to the exclusion of perception, is less a product of foundationalism than of Hume’s aspiration to reduce perceptual knowledge itself to causal inference. This is a special case of an epistemic reductionism that extends to other minds, testimony, and so forth—an imperialism of causal inference, not foundationalism.39 Does the increased reliance on primitive instincts in the Enquiries reflect a desire to prune the psychological and philosophical complexity of the Treatise, or changes in Hume’s view?40 And if changes in view, were they forced upon him? Difficult questions. There is no question that the first Enquiry, taken as a self-contained work, treats perception of body as primitive, thus taking a step in the direction of Reid’s proliferation of first principles. Perhaps these were among the features of the work Hume had in view when, late in life, he declared the Enquiries “a compleat Answer to

39. It might be argued that the position of the Treatise at least betrays foundationalist tendencies. Perhaps; Hume need not have been fully consistent in his rejection of foundationalism. On the other hand, there is room for the view that the theory of ideas was largely an uncritical inheritance, a cluster of pegs on which Hume hung his positive views. Cf. Stroud 1977, 8–27, and Johnson 1995, 37–38. 40. See Kemp Smith 1941, 533–36. For some literature focused on sympathy, see Abramson 2001; Vitz 2004; and Debes 2007.

324

reflection and the stability of belief

Dr Reid and to that bigoted silly Fellow, Beattie” (LDH 2.301). Reid himself would quote approvingly the entire paragraph (EHU 151) where Hume attributes belief in body to natural instinct (EIP 173).

6. Conflicts within the Cognitive Faculties I turn to another dimension—beyond their attitudes toward reductionism and psychological explanation—along which it is instructive to compare the naturalisms of Hume and Reid. Reid maintains that the intellectual faculties, when used properly, do not conflict: “Common sense and reason have both one author; that Almighty author, in all whose other works we observe a consistency, uniformity, and beauty . . . : there must therefore be some order and consistency in the human faculties, as well as in other parts of his workmanship” (IHM 68).41 Though this is a juncture where Reid’s providentialism guides his epistemological construction, as with other claims about the epistemic status of first principles, he need not rely on premises about God. The argument can be understood as inductive: because consistency and uniformity are observed in other parts of nature, they are to be expected in the human faculties.42 The claim that the intellectual faculties do not conflict has a corollary: that the faculties, when properly used, are coequal or coordinate: “The first principles of every kind of reasoning are given us by nature, and are of equal authority with the faculty of reason itself, which is also the gift of Nature” (IHM 172). Reid asks, again applying the consistency of the faculties to reason: “Why . . . should I believe the faculty of reason more than that of perception; they came both out of the same shop, and were made by the same artist . . . ?” (IHM 169). Thus, “when Reason is properly employed, she will confirm the documents of Nature” (IHM 202; cf. EIP 252). The claim that the faculties are consistent is of considerable importance to Reid, and in marked contrast to Hume. Consider the belief in secondary qualities. Hume held that we instinctively believe that bodies possess properties 41. “I apprehend that the word faculty is most properly applied to those powers of the mind which are original and natural, and which make a part of the constitution of the mind” (EIP 21). Clearly, consciousness, perception, memory, et cetera, are faculties. See de Bary 2002, 76–80. 42. The availability of these arguments leaves open the possibility that Reid regards the consistency claim as a first principle. Reid allows that a first principle can be strengthened or confirmed by argument—either by inductive argument (IHM 170–71, 195; EIP 463, 489) or by relying on first principles to establish the existence of a benevolent God (IHM 196). Reid’s first principles include one seeming metaprinciple, “That the natural faculties . . . are not fallacious” (EIP 480). (The role of this principle is controversial. See de Bary 2002, 74–82.) Reid might have thought the no conflict claim a consequence of this principle, though I doubt that it can be made to follow. That the faculties are not fallacious presumably means that they are not systematically mistaken. Natural faculties susceptible even to occasional error would be susceptible to conflict. Reid might have thought consistency of the faculties follows from differences in their domains, but this does not seem apt for testimony, which is subject-matter neutral. Perhaps we should construe the no conflict claim as an implicit first principle, or perhaps it does require argument. These are open questions.

the naturalisms of hume and reid

325

that resemble sensory experiences of color and other secondary qualities (EHU 78n.; cf. T 167, 238–39) and that this belief conflicts with reflection (T 226–27; Essays 235)—in violation of Reid’s strictures in regard to the consistency of the faculties. Hume has no doubt that reflection has things right. Similarly, for Descartes, a spontaneous impulse or inclination to believe in resembling secondary qualities conflicts with clear and distinct perception.43 Descartes, of course, gives the verdict to clear and distinct perception. The problem for Reid is to show that the conflict does not arise. He agrees that there is no resemblance between color experiences and the properties of objects (EIP 85–95). Were the unlearned more or less universally, and at a young age, to believe there is a resemblance, a conflict among faculties would be in the offing. Reid’s solution is to attribute to the common person a causal or dispositional account of color: “By the constitution of our nature, we are led to . . . the conception and belief of some unknown quality in the body, which occasions the idea”; “it is to this quality . . . that we give the name of colour” (IHM 86; cf. 88 and EIP 194, 205). Because the color appearance and its unknown cause “go together in the imagination, and are . . . closely united,” they “are apt to be mistaken for one and the same thing” (IHM 86–87; cf. 89 and EIP 194–95, 204–5). The belief that color experiences resemble properties of objects is a confused accretion to an instinctive belief. By the lights of Descartes and Hume, Reid needs to redescribe—to distort—the content of the common belief in color. In a 1762 letter, Hume puts this disagreement on the record: Reid “supposes, that the Vulgar do not believe the sensible Qualities of Heat, Smell, Sound, & probably Colour to be really in the Bodies, but only their Causes or something capable of producing them in the Mind. But this is imagining the Vulgar to be Philosophers & Corpuscularians from their Infancy” (CTR 18–19).44 The case of the belief in secondary qualities is pallid compared to other conflicts between reason and instinct Hume claims to expose. In the first Enquiry, Hume writes not only that the belief in external body is instinctive (EHU 151) but also that it carries with it the supposition that “the very images, presented by the senses . . . be the external objects” (151). Instinctively, we are not only realists, but direct realists. Reid welcomed these acknowledgments (EIP 173–74), but not the developments to follow. Hume argues that direct realism “is soon destroyed by the slightest philosophy” (EHU 152)—the “diminishing table” argument—so that there is a conflict between instinct and “the obvious dictates of reason” (152).45 This conflict 43. See Meditation VI, paragraphs 14–15; Principles of Philosophy I.66–70. 44. De Bary, 2002 (43–45), has noticed these points and cites this letter in this connection (2002, 47n.21). As de Bary observes, there is some tendency in Intellectual Powers to relax the claim that the common person subscribes to the causal or dispositional form of the belief; rather, the common person does not distinguish this from the belief in its naïve, resemblance form. 45. There is an important moral here. In the first Enquiry, Hume maintains his commitment to the way of ideas—we are directly aware only of perceptions—but allows that belief in the external world is instinctive (§5). That Hume held this position, and that it seems coherent, shows that Reid’s reliance on instinctive first principles can be overlaid on the way of ideas itself. For this sort of reason, disputes related to the way of ideas are distractions from the ways in which Reid is aligned with Hume against Descartes. For a similar line of thought, see Greco 1995, esp. §4.

326

reflection and the stability of belief

is distinctive to the first Enquiry; in the Treatise, the vulgar do not distinguish impressions and external objects (T 193, 202). In both the first Enquiry and the Treatise, realism in any form succumbs to a Berkelian argument: color and other secondary qualities exist only in the mind; we cannot conceive of an object possessing extension and shape separate or abstracted from all sensible secondary qualities; material objects are therefore impossible (226–31; EHU 154–55). Realism is “contrary to reason” (EHU 155; cf. T 231); there is “a direct and total opposition betwixt our reason and our senses” (T 231). Though Hume nonchalantly refers to these arguments as “sceptical topic[s]” (EHU 154), they are not “skeptical” in the way of Cartesian hypotheses: perhaps there is a deceiver; perhaps sense-perception or reason is unreliable. To the thought that our faculties might be unreliable, Reid has his reply that “If we are deceived . . . there is no remedy” (IHM 72). The arguments about realism and direct realism purport to locate conflicts—antinomies—that undermine our faculties. One need not subscribe to a distinctive philosophical theory to want to resist these arguments. Still, they are at odds with Reid’s doctrine that the faculties are consistent and cannot harbor genuine conflict.46 Let us turn to an argument distinctive to the Treatise. In I.iv.1, “Of scepticism with regard to reason,” Hume contends that all demonstration reduces to probability, and that all probability reduces “to nothing, . . . utterly subvert[ing] all belief and opinion” (T 184). (The argument here is distinct from the Russellian problem of induction often attributed to Hume in I.iii.6.) Drawing on this material, Hume announces, “I am ready to reject all belief and reasoning, and can look upon no opinion even as more probable or likely than another” (268–69)—this, provocatively, in I.iv.7, the conclusion of Book I. This is an undermining argument of another sort, purporting to show that a single faculty contains the seeds of its own destruction. Presumably, the consistency, uniformity, and beauty of nature apply to the internal workings of a single faculty. Accordingly, Reid maintains that “first principles . . . will always, from the constitution of human nature, support themselves, and gain rather than lose ground among mankind” (EIP 463).47 If the faculties support themselves, reason cannot self-destruct. The consistency of the faculties leaves no space either for the subversion of reason or for the Humean antinomies. Unsurprisingly, Reid is quite exercised by Hume’s arguments for conflict and subversion. He bemoans “the ignoble attempts of our modern sceptics to depreciate the human understanding, and to dispirit men in the search of truth, by representing the human faculties as fit for nothing, but to lead us into absurdities and contradictions” (IHM 77; cf. EIP 562). Reid frequently alludes to the reduction of probability to zero in I.iv.1 and I.iv.7, where Hume professes readiness to look upon all opinions as equally probable (IHM 3–4, 94; EIP 63, 46. De Bary gives some play to this difference between Reid and Hume (2002, 13, 45, 161–62). 47. For some qualifications, see de Bary 2002, 162–63.

the naturalisms of hume and reid

327

165, 187, 450, 566). Whereas he devotes a scant four pages to Hume in his survey of the history of the way of ideas, much later in Intellectual Powers he devotes a ten-page chapter to criticism of Hume’s argument that reason subverts itself.48 Reid found the undermining arguments disfiguring, so much so as to obscure the project he has in common with Hume—to provide a naturalistic account of the knowledge of all animals, however reflective. Let me offer a subsidiary speculation in regard to the import of Reid’s antipathy for the undermining arguments. Reid quotes Hume’s comment in I.iv.1 (T 183) that “Nature, by an absolute and uncontrollable necessity has determin’d us to judge as well as to breathe and feel” (EIP 571). Similarly, he quotes the paragraph (T 269) that follows Hume’s I.iv.7 announcement that he is “ready to reject all belief and reasoning”: “since reason is incapable of dispelling these clouds, nature herself suffices to that purpose, and cures me of this philosophical melancholy and delirium” (EIP 562). Reid comments: “This was surely a very kind and friendly interposition of Nature. . . . But what pity is it, that Nature . . . so kind in curing this delirium, should be so cruel as to cause it. Doth the same fountain send forth sweet waters and bitter?” (563). Reid is quick to latch on to Hume’s claims of irresistibility and unavoidability in the aftermath of the undermining argument for the subversion of belief.49 Yet, there is little evidence that Reid takes note of the numerous passages that commit Hume to the irresistibility of demonstration, memory, and causal inference. (See §§2–3.) Bear in mind that, apart from one explicit passage in Part iv of Book I, these claims need to be extracted from an assortment of sections in Part iii. There is room for the suggestion that Reid has some tendency to regard the role of irresistibility and unavoidability in Hume as little more than an expedient to extricate us from absurdities and contradictions, not as a component in an independently motivated naturalistic epistemology.

7. Conflict and Stability Reid is also aware that Hume does not always portray instinct as a savior. For Hume, he notes, we are “born under a necessity of believing contradictions” (EIP 562). Hume does construe the Berkelian argument against realism as giving rise to a “direct and total opposition,” a “manifest contradiction,” specifically between perceptual belief in body and causal inference (T 231, 266). These “two 48. The first Enquiry reduces the line of development from I.iv.1 to I.iv.7 to an allusion to “the natural weakness of human understanding” (EHU 158). This provides an additional sense in which this work is an answer to Reid (§5). 49. Reid is explicit that Hume takes the belief in body to be irresistible (IHM 68; EIP 46, 173) and writes that Hume “has been so candid as to acknowledge” (EIP 46; cf. 448) the necessity of trusting the senses and of believing with the vulgar. (Interestingly, Reid cites EHU 151–52 but not Treatise 187 and 218 in I.iv.2. Reid shows no interest in this section, perhaps because it takes the way of ideas for a starting point. Also, in Reid’s view, Berkeley had shown that this leads to skepticism about body; Hume’s contribution was to do away with the self. Cf. IHM 34 and EIP 162.) Reid perhaps sees this candid acknowledgment as Hume’s attempt to salvage belief in the external world in the face of skepticism about body.

328

reflection and the stability of belief

operations,” moreover, are “equally natural and necessary in the human mind” (266). Since there is no way to “adjust those principles together,” philosophers “successively assent to both” (266). In this context, instinctive mechanisms are a source of seemingly ineliminable instability—part of the problem, not the solution. This is a difficulty for the picture of the role of irresistibility in Hume I have attributed to Reid, and also for the Kemp Smith interpretation, in which Hume approves irresistible beliefs.50 Reid writes that Hume “has shewn that [principles commonly received] overturn all knowledge, and at last overturn themselves, and leave the mind in perfect suspense” (EIP 462). This is well off the mark. For Hume, reflection on the antinomies and the subversion of the understanding generates instability in belief—psychological crises, not Pyrrhonian tranquility. (Nonhuman animals are immune to reflection’s conundrums and achieve stability on the cheap.) These crises and their repercussions are part of the subject matter of the science of human nature, as in the closing pages of I.iv.7. In the Treatise, instability in our doxastic lives is one of Hume’s major preoccupations, quite apart from the contexts where the undermining arguments are in play. In Books I and III, reflection on belief-forming mechanisms has the capacity to reinforce or to undermine their use (§5). In an important strand in Part iii of Book I, Hume takes belief to be an infixed and hence a steady disposition (cf. T 86, 109, 118–19, 225, 453, 624, 626, 629/1.3.7.7).51 In I.iii.9, association by relations other than cause and effect is “fluctuating and uncertain,” and “’tis impossible it can ever operate with any considerable degree of force and constancy” (109). In I.iv.2, conflicts with regard to the belief in body involve “combat” (205) and “struggle” between opposing principles “attack’d” by “enemies” which seek to “destroy” (215) one another. In Book II, “Probability arises from an opposition of contrary chances or causes, by which the mind . . . is incessantly tost from one [side] to another. . . . The imagination or understanding . . . fluctuates betwixt the opposite views” (440). In Book III, “continual fluctuation” in judgments of characters and persons gives way to “steady and general points of view,” so that we might “arrive at a more stable judgment” (581–82, Hume’s emphasis). Also in Book III, the mere fact of encountering disagreement with others “disturb[s] the easy course of my thought,” causing “commotion” and “conflict” (592, 593; cf. Essays 60–61). Such texts vindicate an observation of John Passmore’s in 1952: “Associationism comes to be a special example . . . only, of a much more general principle . . . that the mind moves in whatever direction will bring it most ease.”52 50. See my 2002, 23–25. 51. Reid, focusing on Hume’s associationism, overlooks this vein in Hume’s thinking. In fairness, this insight largely had to await interpretive work in the twentieth-century style of analytic reconstruction. Early commentators to spot the dispositional strand were Laird 1932, 88, and then two decades later MacNabb 1951/1966, 77–78, and Basson 1958, ch. 3. This interpretive option is quite clear in Price 1969, esp. 186–87, and Armstrong 1973, 71, and has been taken up in much recent literature. It should be said that Reid’s criticisms of Hume’s official account of belief, in terms of vivacity, are often on the mark. See IHM 30, 197–98, and EIP 291, 401. 52. Passmore makes the point (1952/1968, 122) in the context of his treatment of the conflicts Hume discusses in I.iv.2.

the naturalisms of hume and reid

329

Picking up a thread in D. G. C. MacNabb, writing one year earlier, Hume assigns epistemic pride of place to mechanisms that set us at ease and thus promote stability.53 Suppose Hume was attracted to a theory of justification in this general region. Hume had taken a step beyond Reid, transmuting a naturalistic theory that emphasizes irresistibility and unavoidability into one that places a premium on steadiness and stability. Hume would have thought this step necessary to address problems Reid raises but leaves unresolved. For Reid, the love of simplicity and preference for familiar analogies are “bias[es]” or “prejudice[s]” (EIP 529) that infect the intellectual powers, distinctively human faculties. How are these frailties to be contained? Consider the attention Reid gives to “the bias of human nature . . . to judge from too slight analogies” (EIP 529; cf. IHM 211). This bias accounts for our drawing inferences “rashly” on the basis of accidental conjunctions (IHM 41). But which generalizations are rash? What degrees and kinds of similarity are admissible? Reid’s guidance consists in admonishing Newtonian caution and directing his readers to Bacon’s and Newton’s methodological rules (12; EIP 47–52, 76–87)—though, as I have noted (§2), not to Hume’s version of these rules in I.iii.15. This section contributes to a broader, albeit sketchy, theory of “general rules” (T 149; cf. 110, 146, 173, 631–32/1.3.10.11). In I.iii.13, Hume addresses the question of how to avoid generalizations “we rashly form to ourselves” (146), generalizations based on “accidental” conjunctions (149). Hume’s thought is that when an analogy is too slight, or based on irrelevant similarities, we are prone to conflicting beliefs based on different analogical arguments. The conflicting beliefs motivate us to form generalizations about the extent to which various kinds of analogical argument lead to conflict. The rash generalizations are “of an irregular nature, and destructive of all the most establish’d principles of reasonings” (150); the second-order generalizations are “more extensive and constant” (149) and regulate and control cognitive biases. Conflicts within inductive inference thus lead to greater stability and to improved cognitive performance. In the final sections of the Treatise and the first Enquiry, Hume ups the ante: could we confront the undermining arguments—and the pervasive and fundamental conflicts that attend them—and come out on the other side, with justification intact? In his arguments for the antinomies and subversion of reason, Hume grossly exaggerates the conflicts to which reflection gives rise. Reid’s patient, detailed diagnoses of where these arguments go astray are often on target (e.g., EIP 179–83, 563–72). Even so, Reid’s distaste for conflicts within the faculties would deprive Hume of one of his signature styles of argument. Hume’s project is to show that justification is tied to stability, and that in many cases conflict generates uneasiness that motivates the identification and adoption of salient stabilizing devices. The extreme problem Hume sets himself is to explain how we might reclaim stability and justification even in the face of the undermining arguments. 53. MacNabb 1951/1966, 72–79, 96–100, 191–97. My 2002 develops this line of interpretation and provides supporting detail for many of the points in this section.

330

reflection and the stability of belief

One set of interpretive options, especially well suited to the Treatise, exploits Hume’s insistence that intense reflection cannot be sustained (T 183, 187, 218, 269). Reid notices these sorts of passages and construes them as ad hoc, an admission of epistemological failure (§6). We may suppose instead that, for Hume, justification is relative to stability within a time frame; justification collapses during the short periods of intense reflection but is otherwise preserved. Alternatively, perhaps mechanisms are justifying if they are conducive to sufficient average stability in belief over time. Intense reflection lowers the overall average but not below the threshold of stability required for justification.54 Either way, the intense reflection that generates the undermining arguments, though an epistemic negative, leaves sufficient stability in place to sustain the science of man. A second approach emerges most clearly in the first Enquiry and is of special interest in the context of Reid. Michael Williams has developed the reading, which derives from Passmore, in an especially compelling form.55 The interpretation takes its cue from Hume’s suggestion that skepticism serves a stabilizing function insofar as it tempers or moderates the excesses of superstition (EHU 161–62). Exposure to the “sceptical topic[s]” is thus a net epistemic plus. Reid agrees that the love of simplicity and familiar analogies are “apt to lead us wrong” (EIP 528)—often taking us in the direction of extravagant “hypotheses and systems” (IHM 41). He and Hume share the Newtonian objective of reining in judgments that do not adhere closely to “daily . . . experience” (EHU 162; cf. IHM 125 and EIP 49, 535). How can we achieve the “proper regulation and restraint” (EIP 528) in our reasoning? Hume offers a prescription: though intense reflection gives rise to crises that are temporarily destabilizing, in the longer term “a small tincture of Pyrrhonism” (EHU 161) restrains the imagination from “running, without controul” (162). Our faculties lead us into contradictions but are not “fit for nothing”; the undermining arguments Reid abhors serve a virtuous psychological and epistemic function. That Hume would offer this speculation is another symptom of the centrality of stability in his philosophy.56

54. For discussion and development, see my 2004, 360–62 and 373–76. 55. Passmore 1952/1968, 149–51, and M. Williams 2004, esp. §6. 56. I am grateful to The American Philosophical Association’s Committee on Lectures, Publications, and Research for the honor of selecting me to deliver the 2006–2007 Romanell Lecture. Thanks to the audience for terrific questions—not all answered here. Thanks also to Alexander Broadie and Knud Haakonssen for responding to inquiries, and especially to Stephen Darwall and Frederick Schmitt for help and encouragement. I am indebted to Matthew Davidson, whose questions about an earlier paper—see note 3—prompted me to consider the present topic. Erica Lucast provided superb assistance in verifying selected references and quotations.

the naturalisms of hume and reid

331

This page intentionally left blank

bibliography

For years in the form x/y, x gives the year of the first edition and y the year of the most recent edition consulted. References and citations are from the most recent edition, unless otherwise indicated in the notes. Aaron, Richard I. 1931. “Locke and Berkeley’s Commonplace Book.” Mind 40:439–59. ______. 1937. John Locke. London: Oxford University Press. Abramson, Kate. 1999. “Correcting Our Sentiments about Hume’s Moral Point of View.” Southern Journal of Philosophy 37:333–61. ______. 2001. “Sympathy and the Project of Hume’s Second Enquiry.” Archiv für Geschichte der Philosophie 83:45–80. ______. 2002. “Two Portraits of the Humean Agent.” Pacific Philosophical Quarterly 83:301–34. Adam, Charles. 1910. Vie & Oeuvres de Descartes, Étude Historique. Paris: Léopold Cerf. Alexander, Peter. 1974. “Boyle and Locke on Primary and Secondary Qualities.” Ratio 16:51–67. Alston, William P. 1985. “Thomas Reid on Epistemic Principles.” History of Philosophy Quarterly 2:435–52. Annas, Julia, and Jonathan Barnes. 1985. The Modes of Scepticism, Ancient Texts and Modern Interpretations. Cambridge: Cambridge University Press. Árdal, Páll S. 1966/1989. Passion and Value in Hume’s Treatise. Edinburgh: Edinburgh University Press. Armstrong, David M. 1961. Perception and the Physical World. London: Routledge & Kegan Paul.

333

______. 1973. Belief, Truth and Knowledge. London: Cambridge University Press. Arnold, N. S. 1983. “Hume’s Skepticism about Inductive Inference.” Journal of the History of Philosophy 21:31–55. Atkinson, R. F. 1976. “Hume on the Standard of Morals.” In Kenneth R. Merrill and Robert W. Shahan (eds.). David Hume, Many-Sided Genius, 25–44. Norman: University of Oklahoma Press. Aune, Bruce. 1991. Knowledge of the External World. London: Routledge. Ayer, A. J. 1936. Language, Truth, and Logic. London: Victor Gollancz. ______. 1940. The Foundations of Empirical Knowledge. London: Macmillan. ______. 1963. A Concept of a Person and Other Essays. London: Macmillan. ______. 1980. Hume. New York: Hill and Wang. Ayers, Michael. 1984. “Berkeley and Hume: A Question of Influence.” In Richard Rorty, J. B. Schneewind, and Quentin Skinner (eds.). Philosophy in History, Essays on the Historiography of Philosophy, 303–27. Cambridge: Cambridge University Press. ______. 1991. Locke, vol. 1, Epistemology. London: Routledge. ______. 1994. “The Foundations of Knowledge and the Logic of Substance: The Structure of Locke’s General Philosophy.” In G. A. J. Rogers (ed.). Locke’s Philosophy, Content and Context, 49–73. Oxford: Clarendon. ______. 2005. “Was Berkeley an Empiricist or a Rationalist?” In Kenneth P. Winkler (ed.). The Cambridge Companion to Berkeley, 34–62. Cambridge: Cambridge University Press. Baier, Annette. 1982. “Hume’s Account of Our Absurd Passions.” Journal of Philosophy 79:643–51. ______. 1991. A Progress of Sentiments, Reflections on Hume’s Treatise. Cambridge, Massachusetts: Harvard University Press. Barnes, Jonathan. 1982. “The Beliefs of a Pyrrhonist.” Proceedings of the Cambridge Philological Society n.s. 28:1–29. Basson, A. H. 1958. David Hume. Harmondsworth, England: Penguin. (Reprinted, New York: Dover, 1969, as David Hume, by A. P. Cavendish.) Beauchamp, T. L., and T. A. Mappes. 1975. “Is Hume Really a Sceptic about Induction?” American Philosophical Quarterly 12:119–29. Beauchamp, Tom L., and Alexander Rosenberg. 1981. Hume and the Problem of Causation. New York: Oxford University Press. Beck, L. J. 1952. The Method of Descartes, A Study of the Regulae. Oxford: Clarendon. ______. 1965. The Metaphysics of Descartes, A Study of the Meditations. Oxford: Clarendon. Bennett, Jonathan. 1971. Locke, Berkeley, Hume: Central Themes. Oxford: Clarendon. ______. 1977. “Descartes’ Theory of Modality.” Philosophical Review 103: 639–67. ______. 1987. “Substratum.” History of Philosophy Quarterly 4:197–215. ______. 1990. “Truth and Stability in Descartes’s Meditations.” Canadian Journal of Philosophy, supplementary volume 16:75–108. ______. 2001. Learning from Six Philosophers, two vols. Oxford: Clarendon.

334

bibliography

Blackburn, Simon. 1990. “Hume and Thick Connexions.” Philosophy and Phenomenological Research, supplementary volume 50:327–50. Blanshard, Brand. 1925. Review of Prolegomena to an Idealist Theory of Knowledge, by Norman Kemp Smith. Journal of Philosophy 22: 605–9. Bracken, Harry M. 1958. The Early Reception of Berkeley’s Immaterialism. The Hague: Martinus Nijhoff. ______. 1974. Berkeley. New York: St. Martin’s Press. Brand, Walter. 1992. Hume’s Theory of Moral Judgment: A Study in the Unity of A Treatise of Human Nature. Dordrecht: Kluwer. Bréhier, Emile. 1937. “La Création des vérités éternelles.” Revue Philosophique de la France et de l’Étranger 113:15–29. Page references to the republication as “The Creation of the Eternal Truths in Descartes’ System.” In Willis Doney (ed.), 1967. Descartes, a Collection of Critical Essays, 192–208. Garden City, New York: Anchor. Bricke, John. 1980. Hume’s Philosophy of Mind. Princeton, New Jersey: Princeton University Press. ______. 1996. Mind and Morality: An Examination of Hume’s Moral Psychology. Oxford: Clarendon. Broackes, Justin. 1993. “Does Hume Hold a Regularity Theory of Causation?” British Journal for the History of Philosophy 1:99–114. Broad, C. D. 1914. Perception, Physics and Reality. Cambridge: Cambridge University Press. ______. 1923. Scientific Thought. London: Kegan Paul, Trench, Trubner. ______. 1930. Five Types of Ethical Theory. London: Routledge and Kegan Paul. ______. 1942. “Certain Features in Moore’s Ethical Doctrines.” In Paul Arthur Schilpp (ed.). The Philosophy of G. E. Moore, 43–67. Evanston, Illinois: Northwestern University. ______. 1953. “Self and Others.” Herbert Spencer Lecture, Oxford University. As first published in David R. Cheney (ed.), 1971. Broad’s Critical Essays in Moral Philosophy, 262–82. London: George Allen and Unwin. Broughton, Janet. 1983. “Hume’s Skepticism about Causal Inferences.” Pacific Philosophical Quarterly 64:3–18. ______. 1987. “Necessity and Physical Laws in Descartes’s Philosophy.” Pacific Philosophical Quarterly 68:205–21. ______. 2008. “Hume’s Explanation of Causal Inference.” In Paul Hoffman, David Owen, and Gideon Yaffe (eds.). Contemporary Perspectives on Early Modern Philosophy: Essays in Honor of Vere Chappell, 289–305. Toronto: Broadview. Brown, Charlotte. 1994. “From Spectator to Agent: Hume’s Theory of Obligation.” Hume Studies 20:19–35. ______. 2001. “Is the General Point of View the Moral Point of View?” Philosophy and Phenomenological Research 62:197–20. Buchdahl, Gerd. 1969. Metaphysics and the Philosophy of Science: The Classical Origins, Descartes to Kant. Oxford: Oxford University Press.

bibliography

335

Burnyeat, Myles F. 1980. “Can the Skeptic Live His Skepticism?” In Malcolm Schofield, Myles Burnyeat, and Jonathan Barnes (eds.). Doubt and Dogmatism, Studies in Hellenistic Philosophy, 20–53. Oxford: Clarendon. _____. 1984. “The Sceptic in His Place and Time.” In Richard Rorty, J. B. Schneewind, and Quentin Skinner (eds.). Philosophy in History, Essays on the Historiography of Philosophy, 225–54. Cambridge: Cambridge University Press. Butler, Ronald J. 1960. “Natural Belief and the Enigma of Hume.” Archiv für Geschichte der Philosophie 42:73–100. Carruthers, Peter. 1990. “What Is Empiricism?” Aristotelian Society, supplementary volume 64:63–79. Caton, Hiram. 1970. “On the Interpretation of the Meditations.” Man and World 3:224–45. ______. 1971. “The Problem of Descartes’ Sincerity.” Philosophical Forum 2: 355–70. ______. 1973. The Origin of Subjectivity: An Essay on Descartes. New Haven, Connecticut: Yale University Press. ______. 1975. “Will and Reason in Descartes’ Theory of Error.” Journal of Philosophy 72:87–104. Chappell, V. C. 1994. “Locke’s Theory of Ideas.” In Vere Chappell (ed.). The Cambridge Companion to Locke, 26–55. Cambridge: Cambridge University Press. Chappell, Vere, G. A. J. Rogers, Gideon Yaffe, Lex Newman, Martha Brandt Bolton, and Thomas M. Lennon. 2004. “Symposium: Locke and the Veil of Perception.” Pacific Philosophical Quarterly 85:243–363. Chisholm, Roderick. 1948. “The Problem of Empiricism.” Journal of Philosophy 45:412–17. Christensen, David. 1994. “Conservatism in Epistemology.” Noûs 28:69–89. Cohen, Mendel. 1990. “Obligation and Human Nature in Hume’s Philosophy.” Philosophical Quarterly 40:316–41. Cohen, Stewart. 2002. “Basic Knowledge and the Problem of Easy Knowledge.” Philosophy and Phenomenological Research 65:309–29. Cohon, Rachel. 1997. “The Common Point of View in Hume’s Ethics.” Philosophy and Phenomenological Research 57:827–50. Connon, R. W. 1976. “The Naturalism of Hume Revisited.” In David Fate Norton, Nicholas Capaldi, and Wade L. Robison (eds.). McGill Hume Studies, vol. 1, Studies in Hume and Scottish Philosophy, 121–45. San Diego: Austin Hill. Cooper, John M. 1970. “Plato on Sense-Perception and Knowledge (Theaetetus 184–186).” Phronesis 15:123–46. Costa, Michael A. 1981. “Hume and Justified Belief.” Canadian Journal of Philosophy 11:219–28. Cottingham, John. 1986. Descartes. Oxford: Blackwell. _____. 1991. “The Ethics of Self-Concern.” Ethics 101:798–817. Craig, Edward. 1987. The Mind of God and the Works of Man. Oxford: Oxford University Press. _____. 1990. Review of Thomas Reid, by Keith Lehrer. Ratio n.s. 3:182–85.

336

bibliography

Cress, Donald A. (trans.). 1979. René Descartes, Meditations on First Philosophy. Indianapolis, Indiana: Hackett. Curley, E. M. 1978. Descartes against the Skeptics. Cambridge, Massachusetts: Harvard University Press. ______. 1984. “Descartes on the Creation of the Eternal Truths.” Philosophical Review 93:569–97. ______ (ed. and trans.). 1985. The Collected Works of Spinoza. Princeton, New Jersey: Princeton University Press. Dalrymple, Houghton. 1982. “Skeptical Tendencies in Hume.” Southwest Philosophical Studies 7:75–82. ______. 1986. “Kemp Smith, Hume and the Parallelism between Reason and Morality.” Hume Studies 12:77–91. Daniels, Norman. 1976. “On Having Concepts ‘by Our Constitution.’” In Stephen Barker and Tom L. Beauchamp (eds.). Thomas Reid, Critical Interpretations, 35–43. Philadelphia: Temple University Press. Darwall, Stephen. 1986. “Agent-Centered Restrictions from the Inside Out.” Philosophical Studies 50:291–319. ______. 1995. “Hume and the Invention of Utilitarianism.” In M. A. Stewart and John P. Wright (eds.). Hume and Hume’s Connexions, 58–92. University Park: Pennsylvania State University Press. ______. 1998. “Empathy, Sympathy, Care.” Philosophical Studies 89:261–82. ______. 1999. “Sympathetic Liberalism: Recent Work on Adam Smith.” Philosophy and Public Affairs 28:139–64. Dauer, Francis W. 1980. “Hume’s Skeptical Solution and the Causal Theory of Knowledge.” Philosophical Review 89:357–78. Davidson, Arnold I., and Norbert Hornstein. 1984. “The Primary/Secondary Quality Distinction: Berkeley, Locke, and the Foundations of Corpuscularian Science.” Dialogue 23:281–303. de Bary, Philip. 2002. Thomas Reid and Scepticism: His Reliabilist Response. London: Routledge. Debes, Remy. 2007. “Has Anything Changed? Hume’s Theory of Association and Sympathy after the Treatise.” British Journal for the History of Philosophy 15:313–38. Della Rocca, Michael. 1999. “‘If a Body Meet a Body’: Descartes on Body-Body Causation.” In Rocco Gennaro and Charles Huenemann (eds.). New Essays on the Rationalists, 48–81. New York: Oxford University Press. ______ 2005. “Descartes, the Cartesian Circle, and Epistemology without God.” Philosophy and Phenomenological Research 70:1–33. DeRose, Keith. 1989. “Reid’s Anti-Sensationalism and His Realism.” Philosophical Review 98:313–48. Doney, Willis. 1955. “The Cartesian Circle.” Journal of the History of Ideas 16:324–38. ______. 1970. “Descartes’s Conception of Perfect Knowledge.” Journal for the History of Philosophy 8:387–403. Dorter, Kenneth. 1973. “Science and Religion in Descartes’ Meditations.” Thomist 37:313–40.

bibliography

337

Dutton, Blake D. 1996. “Indifference, Necessity, and Descartes’s Derivation of the Laws of Motion.” Journal of the History of Philosophy 34:193–212. Egan, Louisa, Laurie R. Santos, and Paul Bloom. 2007. “The Origins of Cognitive Dissonance.” Psychological Science 18:978–83. Etchemendy, John. 1981. “The Cartesian Circle: Circulus ex Tempore.” Studia Cartesiana 2:5–42. Everson, Stephen. 1988. “The Difference between Feeling and Thinking.” Mind 97:401–13. Ewing, A. C. 1942. Critical notice of The Philosophy of David Hume, by Norman Kemp Smith. Mind 51:69–75. ______. 1959. “Norman Kemp Smith, 1872–1958.” Proceedings of the British Academy 45:297–306. Ewing, A. C., R. I. Aaron, D. G. C. MacNabb, and H. H. Price. 1945. “Symposium: The Causal Argument for Physical Objects.” Aristotelian Society, supplementary volume 19:32–100. Feldman, Fred, and Arnold Levison. 1971. “Anthony Kenny and the Cartesian Circle.” Journal of the History of Philosophy 9:491–96. Firth, Roderick. 1964. “Coherence, Certainty, and Epistemic Priority.” Journal of Philosophy 19:545–57. Flage, Daniel E. 1990. Hume’s Theory of Mind. London: Routledge. Flew, Antony. 1961. Hume’s Philosophy of Belief, a Study of his First Inquiry. London: Routledge & Kegan Paul. ______. 1986. David Hume: Philosopher of Moral Science. Oxford: Basil Blackwell. Fogelin, Robert J. 1985. Hume’s Skepticism in the Treatise of Human Nature. London: Routledge & Kegan Paul. ______. 1993. “Hume’s Scepticism.” In David Fate Norton (ed.). The Cambridge Companion to Hume, 90–116. Cambridge: Cambridge University Press. ______. 1998. “Garrett on the Consistency of Hume’s Philosophy.” Hume Studies 24:161–69. Frankfurt, Harry G. 1958. “Peirce’s Account of Inquiry.” Journal of Philosophy 55:588–92. ______. 1962. “Memory and the Cartesian Circle.” Philosophical Review 71:504–11. ______. 1965. “Descartes’ Validation of Reason.” American Philosophical Quarterly 2:146–56. ______. 1970. Demons, Dreamers, and Madmen: The Defense of Reason in Descartes’s Meditations. Indianapolis, Indiana: Bobbs-Merrill. ______. 1977. “Descartes on the Creation of the Eternal Truths.” Philosophical Review 86:36–57. ______. 1978. “Descartes on the Consistency of Reason.” In Michael Hooker (ed.). Descartes: Critical and Interpretive Essays, 26–39. Baltimore: Johns Hopkins University Press. Garber, Daniel. 1986. “Semel in vita: The Scientific Background to Descartes’ Meditations.” In Amélie O. Rorty (ed.). Essays on Descartes’s Meditations, 81–116. Berkeley: University of California Press. Gardner, Sebastian. 2007. “The Limits of Naturalism and the Metaphysics of German Idealism.” In Espen Hammer (ed.). German Idealism, Contemporary Perspectives, 19–49. London: Routledge.

338

bibliography

Garns, Rudy L. 1988. “Descartes and Indubitability.” Southern Journal of Philosophy 26:83–100. Garrett, Don. 1997. Cognition and Commitment in Hume’s Philosophy. New York: Oxford University Press. ______. 1998. “Ideas, Reason, and Skepticism: Response to My Critics.” Hume Studies 24:171–94. ______. 2001. “Replies.” Philosophy and Phenomenological Research 62: 205–18. ______. 2002. “Hume on Testimony concerning Miracles.” In Peter Millican (ed.). Reading Hume on Human Understanding, 301–34. Oxford: Clarendon. ______. 2005. Introduction to The Philosophy of David Hume, by Norman Kemp Smith. Houndmills, England: Palgrave Macmillan. Gaukroger, Stephen. 1995. Descartes, an Intellectual Biography. Oxford: Clarendon. Gewirth, Alan. 1941. “The Cartesian Circle.” Philosophical Review 50:368–95. ______. 1970. “The Cartesian Circle Reconsidered.” Journal of Philosophy 67:668–85. ______. 1971. “Descartes: Two Disputed Questions.” Journal of Philosophy 68:288–96. Gibson, A. Boyce. 1932. The Philosophy of Descartes. London: Methuen. Gibson, James. 1917. Locke’s Theory of Knowledge and Its Historical Relations. Cambridge: Cambridge University Press. Glathe, Alfred B. 1950. Hume’s Theory of the Passions and of Morals. University of California Publications in Philosophy 24. Berkeley: University of California Press. Goldman, Alvin I. 1967. “A Causal Theory of Knowing.” Journal of Philosophy 64:357–72. ______. 1979. “What Is Justified Belief?” In George S. Pappas (ed.). Justification and Knowledge, 1–23. Dordrecht, Holland: D. Reidel. ______. 2004. “Sosa on Reflective Knowledge and Virtue Perspectivism.” In John Greco (ed.). Ernest Sosa and His Critics, 86–95. Malden, Massachusetts: Blackwell. Goldman, Holly Smith. 1976. “Dated Rightness and Moral Obligation.” Philosophical Review 85:449–87. Gombay, André. 1979. “Mental Conflict: Descartes.” Philosophy 54:485–500. Gordon, Robert M. 1995. “Sympathy, Simulation, and the Impartial Spectator.” Ethics 105:727–42. Goudge, Thomas A. 1950. The Thought of C. S. Peirce. Toronto: University of Toronto Press. Gouhier, Henri. 1962. La Pensée Métaphysique de Descartes. Paris: J. Vrin. Greco, John. 1995. “Reid’s Critique of Berkeley and Hume: What’s the Big Idea?” Philosophy and Phenomenological Research 55:279–96. ______. 2004. “Reid’s Reply to the Skeptic.” In Terence Cuneo and René van Woudenberg (eds.). The Cambridge Companion to Thomas Reid, 143–55. Cambridge: Cambridge University Press. Green, Thomas Hill. 1886. “Introduction to the Moral Part of the Treatise.” In Thomas Hill Green and Thomas Hodge Grose (eds.). David Hume, the Philosophical Works, vol. 2, 1–71. London: Longman Green.

bibliography

339

Haack, Susan. 1982. “Descartes, Peirce and the Cognitive Community.” Monist 65:156–81. Hacking, Ian. 1975. The Emergence of Probability. Cambridge: Cambridge University Press. ______. 1978. “Hume’s Species of Probability.” Philosophical Studies 33: 21–37. Haldane, Elizabeth S. 1905. Descartes, His Life and Times. London: John Murray. Hamilton, Andy. 1998. “Mill, Phenomenalism, and the Self.” In John Skorupski (ed.). The Cambridge Companion to Mill, 138–75. Cambridge: Cambridge University Press. Harman, Gilbert. 1986. “Moral Agent and Impartial Spectator.” The Lindley Lecture. Lawrence: Department of Philosophy, University of Kansas. ______. 2001. “General Foundations versus Rational Insight.” Philosophy and Phenomenological Research 63:657–63. ______. 2003. “Skepticism and Foundations.” In Steven Luper (ed.). The Skeptics: Contemporary Essays, 1–11. Aldershot, England: Ashgate. ______. 2010. “Gilbert Harman.” In Jonathan Dancy, Ernest Sosa, and Matthias Steup (eds.). A Companion to Epistemology, 2nd ed., 152–56. Oxford: Blackwell. Harrison, Jonathan. 1976. Hume’s Moral Epistemology. Oxford: Clarendon. ______. 1981. Hume’s Theory of Justice. Oxford: Clarendon. Hatfield, Gary. 1986. “The Senses and the Fleshless Eye: The Meditations as Cognitive Exercises.” In Amélie O. Rorty (ed.). Essays on Descartes’s Meditations, 45–79. Berkeley: University of California Press. ______. 1993. “Reason, Nature, and God in Descartes.” In Stephen Voss (ed.). Essays on the Philosophy and Science of René Descartes, 259–87. New York: Oxford University Press. ______. 2008. “Animals.” In Janet Broughton and John Carriero (eds.). A Companion to Descartes, 404–25. Malden, Massachusetts: Blackwell. Hearn, Thomas K. 1970. “‘General Rules’ in Hume’s Treatise.” Journal of the History of Philosophy 8:405–22. ______. 1976. “General Rules and the Moral Sentiments in Hume’s Treatise.” Review of Metaphysics 30:57–72 Heffernan, George (ed. and trans.). 1990. Meditationes de prima philosophia, Meditations on First Philosophy, a Bilingual Edition. Notre Dame, Indiana: University of Notre Dame. Hempel, Carl G. 1965. “Empiricist Criteria of Cognitive Significance: Problems and Changes.” In Carl G. Hempel. Aspects of Scientific Explanation and Other Essays in the Philosophy of Science, 101–19. London: Free Press. Hendel, Charles W. 1925 and 1963. Studies in the Philosophy of David Hume. Princeton, New Jersey: Princeton University Press (1925). Indianapolis, Indiana: Bobbs-Merrill (1963). Hicks, G. Dawes. 1932. Berkeley. London: Ernst Benn. Hoffman, Martin. 2000. Empathy and Moral Development: Implications for Caring and Justice. Cambridge: Cambridge University Press. Holland, A. J. 1973. “An Argument in Plato’s Theaetetus: 184–186.” Philosophical Quarterly 91:97–116.

340

bibliography

Hookway, Christopher. 1990a. Scepticism. London: Routledge. ______. 1990b. “Critical Common-Sensism and Rational Self-Control.” Noûs 24:397–412. Howson, Colin. 2000. Hume’s Problem: Induction and the Justification of Belief. Oxford: Clarendon. Humberstone, I. L. 1992. “Direction of Fit.” Mind 101:59–83. Immerwahr, John. 1977. “The Failure of Hume’s Treatise.” Hume Studies 3:57–71. ______. 1979. “A Skeptic’s Progress: Hume’s Preference for the First Enquiry.” In David Fate Norton, Nicholas Capaldi, and Wade L. Robison (eds.). McGill Hume Studies, vol. 1, Studies in Hume and Scottish Philosophy, 227–38. San Diego: Austin Hill. ______. 1992. “Hume on Tranquilizing the Passions.” Hume Studies 18: 293–314. Jackson, Frank. 1977. Perception, a Representative Theory. Cambridge: Cambridge University Press. Jacovides, Michael. 2002. “The Epistemology under Locke’s Corpuscularianism.” Archiv für Geschichte der Philosophie 84:161–89. James, William. 1904a. “Does ‘Consciousness’ Exist?” Journal of Philosophy, Psychology and Scientific Methods 1:477–91. ______. 1904b. “A World of Pure Experience.” Journal of Philosophy, Psychology and Scientific Methods 1:533–43. Jessop, T. E. 1938. “Malebranche and Berkeley.” Revue Internationale de Philosophie 1:121–42. ______. 1948. Review of The Philosophy of David Hume, by Norman Kemp Smith. Philosophy 23:264–68. ______. 1992. “Hume on Tranquillizing the Passions.” Hume Studies 2: 293–314. Johnson, Oliver A. 1995. The Mind of David Hume: A Companion to Book I of A Treatise of Human Nature. Urbana: University of Illinois Press. Jolley, Nicholas. 1990. “Berkeley and Malebranche on Causality and Volition.” In J. A. Cover and Mark Kulstad (eds.). Central Themes in Early Modern Philosophy, 227–44. Indianapolis, Indiana: Hackett. ______. 1996. “Berkeley, Malebranche, and Vision in God.” Journal of the History of Philosophy 34:535–48. Jones, Peter. 1976. “Cause, Reason, and Objectivity in Hume’s Aesthetics.” In Donald W. Livingston and James T. King (eds.). Hume: A Re-evaluation, 323–42. New York: Fordham University Press. Kagan, Shelly. 1989. The Limits of Morality. Oxford: Clarendon. Kahneman, Daniel, Paul Slovic, and Amos Tversky (eds.). 1982. Judgment under Uncertainty: Heuristics and Biases. Cambridge: Cambridge University Press. Kail, P. J. E. 2001. “Projection and Necessity in Hume.” European Journal of Philosophy 9:24–54. Kaufman, Dan. 2005. “God’s Immutability and the Necessity of Descartes’s Eternal Truths.” Journal of the History of Philosophy 43:1–19. Keeling, S. V. 1934/1968. Descartes. London: Oxford University Press. Kemp Smith, Norman. 1902. Studies in the Cartesian Philosophy. New York: Russell & Russell.

bibliography

341

______. 1905. “The Naturalism of Hume (I)” and “The Naturalism of Hume (II).” Mind 14:149–73 and 335–47. ______. 1908. “Subjectivism and Realism in Modern Philosophy.” Philosophical Review 17:138–48. ______. 1912. “How Far Is Agreement Possible in Philosophy?” Journal of Philosophy, Psychology, and Scientific Methods 9:701–11. ______. 1915. “Kant’s Relation to Hume and Leibniz.” Philosophical Review 24:288–96. ______. 1918. A Commentary to Kant’s Critique of Pure Reason. London: Macmillan. ______. 1919. “The Present Situation in Philosophy.” Edinburgh: James Thin. Page references to the republication in Philosophical Review 29 (1920): 1–26. ______. 1924. Prolegomena to an Idealist Theory of Knowledge. London: Macmillan. ______ (ed.). 1935. David Hume, Dialogues concerning Natural Religion. London: Thomas Nelson. ______. 1941. The Philosophy of David Hume: A Critical Study of Its Origins and Central Doctrines. London: Macmillan. ______. 1952. New Studies in the Philosophy of Descartes: Descartes as Pioneer. London: Macmillan. ______. 1981. The Letters of Baron Friedrich von Hügel and Professor Norman Kemp Smith. Lawrence F. Barmann (ed.). New York: Fordham University Press. Kenny, Anthony. 1968. Descartes, a Study of His Philosophy. New York: Random House. ______. 1971. “A Reply to Feldman and Levison.” Journal of the History of Philosophy 9:497–98. Kobes, Bernard W. 1992. “Believers Prefer Truth.” American Philosophical Association, Pacific Division Meetings, Portland, Oregon, March 27. Kornblith, Hilary. 2004. “Sosa on Human and Animal Knowledge.” In John Greco (ed.). Ernest Sosa and His Critics, 126–34. Malden, Massachusetts: Blackwell. Korsgaard, Christine M. 1989. “Normativity as Reflexivity: Hume’s Practical Justification of Morality.” Sixteenth Hume Society Conference, Lancaster, England. ______. 1996. The Sources of Normativity. Onora O’Neill (ed.). Cambridge: Cambridge University Press. ______. 1999. “The General Point of View: Love and Moral Approval in Hume’s Ethics.” Hume Studies 25:3–41. Koyré, Alexandre. 1922. Essai sur l’idée de Dieu et les preuves de son existence chez Descartes. Paris: E. Leroux. La Croix, Richard R. 1984. “Descartes on God’s Ability to Do the Logically Impossible.” Canadian Journal of Philosophy 14:455–76. Laird, John. 1932. Hume’s Philosophy of Human Nature. London: Methuen. ______. 1941. Review of The Philosophy of David Hume, by Norman Kemp Smith. University of Edinburgh Journal 11:129–30.

342

bibliography

Lanzetta, John T., and Basil G. Englis. 1989. “Expectations of Cooperation and Competition and Their Effects on Observers’ Vicarious Emotional Responses.” Journal of Personality and Social Psychology 56: 543–54. Laporte, Jean. 1950. Le Rationalisme de Descartes. Paris: Presses Universitaires de France. Larmore, Charles. 1984. “Descartes’ Psychologistic Theory of Assent.” History of Philosophy Quarterly 1:61–74. Lehrer, Keith. Knowledge. 1974. Oxford: Clarendon. Lennon, Thomas M. 1993. The Battle of the Gods and Giants, the Legacies of Descartes and Gassendi, 1655–1715. Princeton, New Jersey: Princeton University Press. Lenz, John. 1958. “Hume’s Defense of Causal Inference.” Journal of the History of Ideas 19:559–67. Leon, Mark. 1992. “Rationalising Belief.” Philosophical Papers 21:299–314. Leroy, Maxime. 1929. Descartes, le Philosophe au Masque, two vols. Paris: Les Éditions Rieder. Levett, M. J. 1937. “Note on the Alleged Cartesian Circle.” Mind 46:206–13. Lewis, C. I. 1946. An Analysis of Knowledge and Valuation. La Salle, Illinois: Open Court. Lewis, David. 1973. Counterfactuals. Cambridge, Massachusetts: Harvard University Press. Livingston, Donald W. 1984. Hume’s Philosophy of Common Life. Chicago: University of Chicago Press. Loeb, Louis E. 1977a. “Hume’s Moral Sentiments and the Structure of the Treatise.” Journal of the History of Philosophy 15:395–403. ______. 1977b. Review of Why Does Language Matter to Philosophy? by Ian Hacking. Philosophical Review 86:437–40. ______. 1981a. From Descartes to Hume: Continental Metaphysics and the Development of Modern Philosophy. Ithaca, New York: Cornell University Press. ______. 1981b. Review of Hume’s Philosophy of Religion, by J. C. A. Gaskin. Philosophical Review 90:283–85. ______. 1985. “Is There a Problem of Cartesian Interaction?” Journal of the History of Philosophy 23:227–31. ______. 1986. “Is There Radical Dissimulation in Descartes’ Meditations?” In Amélie O. Rorty (ed.). Essays on Descartes’s Meditations, 243–70. Berkeley: University of California Press. This volume, ch. 1. ______. 1988. “Was Descartes Sincere in His Appeal to the Light of Nature?” Journal of the History of Philosophy 26:377–406. ______. 1990. “The Priority of Reason in Descartes.” Philosophical Review 99:3–43. This volume, ch. 2. ______. 1991. “Stability, Justification, and Hume’s Propensity to Attribute Identity to Related Objects.” Philosophical Studies 19:237–70. ______. 1992a. “The Cartesian Circle.” In John Cottingham (ed.). The Cambridge Companion to Descartes, 200–235. Cambridge: Cambridge University Press. This volume, ch. 3.

bibliography

343

______. 1992b. “Causation, Extrinsic Relations, and Hume’s Second Thoughts about Personal Identity.” Hume Studies 18:219–31. ______. 1994. Review of A Progress of Sentiments: Reflections on Hume’s Treatise, by Annette C. Baier. Philosophy and Phenomenological Research 54:467–74. ______. 1995a. “Hume on Stability, Justification, and Unphilosophical Probability.” Journal of the History of Philosophy 33:91–121. This volume, ch. 7. ______. 1995b. “Instability and Uneasiness in Hume’s Theories of Belief and Justification.” British Journal for the History of Philosophy 3:301–29. ______. 1995c. Review of Thought’s Ego in Augustine and Descartes, by Gareth B. Matthews. Mind 104:182–86. ______. 1997. “Causal Inference, Associationism, and Skepticism in Part III of Book I of Hume’s Treatise.” In Patricia A. Easton (ed.). Logic and the Workings of the Mind: The Logic of Ideas and Faculty Psychology in Early Modern Philosophy, 283–305. North American Kant Society Studies in Philosophy, vol. 5. Atascadero, California: Ridgeview. ______. 1998. “Sextus, Descartes, Hume, and Peirce: On Securing Settled Doxastic States.” Noûs 32:205–30. This volume, ch. 4. ______. 2001a. “Hume’s Explanations of Meaningless Beliefs.” Philosophical Quarterly 51:145–64. This volume, ch. 6. ______. 2001b. “Integrating Hume’s Accounts of Belief and Justification.” Philosophy and Phenomenological Research 63:279–303. This volume, ch. 5. ______. 2002. Stability and Justification in Hume’s Treatise. New York: Oxford University Press. ______. 2003. “Hume’s Agent-centered Sentimentalism.” Philosophical Topics 31:309–41. This volume, ch. 8. ______. 2004. “Stability and Justification in Hume’s Treatise—Another Look.” Hume Studies 30:339–404. ______. 2005. “The Mind-Body Union, Interaction, and Subsumption.” In Christia Mercer and Eileen O’Neill (eds.). Early Modern Philosophy, Mind, Matter, and Metaphysics, 65–85. New York: Oxford University Press. ______. 2006. “Psychology, Epistemology, and Skepticism in Hume’s Argument about Induction.” Synthese 152:321–38. This volume, ch. 10. ______. 2007. “The Naturalisms of Hume and Reid.” Proceedings and Addresses of the American Philosophical Association 81:65–92. This volume, ch. 12. ______. 2008. “Inductive Inference in Hume’s Philosophy.” In Elizabeth Radcliffe (ed.). Blackwell Companion to Hume, 106–25. Oxford: Blackwell. ______. 2009a. “What Is Worth Preserving in the Kemp Smith Interpretation of Hume?” British Journal of the History of Philosophy 17:769–97. This volume, ch. 9. ______. 2009b. Review of Projection and Realism in Hume’s Philosophy, by P. J. E. Kail. Mind 118:181–85. ______. Forthcoming. “Locke and British Empiricism.” In Matthew Stuart (ed.). A Companion to Locke. Oxford: Blackwell. This volume, ch. 11. Lovejoy, Arthur O. 1925. Review of Prolegomena to an Idealist Theory of Knowledge, by Norman Kemp Smith. Philosophical Review 34:185–93.

344

bibliography

Luce, A. A. 1934. Berkeley and Malebranche: A Study in the Origins of Berkeley’s Thought. Oxford: Oxford University Press. ______ (ed.). 1944. George Berkeley, Philosophical Notebooks, Generally Called the Commonplace Book. London: Thomas Nelson. ______. 1963. The Dialectic of Immaterialism. London: Hodder & Stoughton. Lynch, Michael P. 1996. “Hume and the Limits of Reason.” Hume Studies 22:89–104. Mach, Ernst. 1886. Die Analyse der Empfindungen und das Verhältnis des Physischen zum Psychischen. Jena: G. Fischer. Fifth edition (1906), Sydney Waterlow (trans.). The Analysis of Sensations and the Relation of the Physical to the Psychical. New York: Dover, 1959. Mackenzie, J. S. 1924. Critical Notice of Prolegomena to an Idealist Theory of Knowledge, by Norman Kemp Smith. Mind 33:322–27. Mackie, J. L. 1976. Problems from Locke. Oxford: Clarendon. ______. 1980. Hume’s Moral Theory. London: Routledge and Kegan Paul. Maclennan, R. D. 1967. “Divine Existence: A Personal Recollection.” In A. J. D. Porteous, R. D. Maclennan, and G. E. Davie (eds.). Norman Kemp Smith, the Credibility of Divine Existence, 38–60. New York: St. Martin’s Press. MacNabb, D. G. C. 1951/1966. David Hume, His Theory of Knowledge and Morality. Oxford: Basil Blackwell. New York: Hutchenson’s University Library (1951). Hamden, Connecticut: Archon (1966). Mahoney, Michael Sean (trans.). 1979. René Descartes, Le Monde, ou Traité de la Lumière. New York: Abaris. Mandelbaum, Maurice. 1964. Philosophy, Science, and Sense-Perception. Baltimore: Johns Hopkins University Press. Markie, Peter J. 1986. Descartes’s Gambit. Ithaca, New York: Cornell University Press. Mawson, T. J. 2001. “Eternal Truths and Cartesian Circularity.” British Journal for the History of Philosophy 9:197–220. McCracken, Charles J. 1983. Malebranche and British Philosophy. Oxford: Clarendon. ______. 1988. “Berkeley’s Concept of Mind: The Return through Malebranche and Locke to Descartes.” Monist 71:596–611. Meeker, Kevin. 1998. “Hume: Radical Skeptic or Naturalized Epistemologist?” Hume Studies 24:31–52. Menn, Stephen. 1998. Descartes and Augustine. Cambridge: Cambridge University Press. Merrylees, W. A. 1934. Descartes, an Examination of Some Features of His Metaphysics and Method. Melbourne: Melbourne University Press, Michael, Emily. 1999. “Reid’s Critique of the Scottish Logic of Ideas.” Reid Studies 2:3–17. Michael, Fred S., and Emily Michael. 1987. “Reid’s Hume: Remarks on Hume in Some Early Logic Lectures of Reid.” Monist 70:508–26. Mijuskovic, Ben. 1971. “Hume and Shaftsbury on the Self.” Philosophical Quarterly 21:324–36. Miller, Alexander. 1994. “Truth, Permanence, and the Regulation of Belief: Loeb’s Cartesian Argument.” Ratio n.s. 7:111–21. Miller, Valentine Rodger, and Reese P. Miller (trans.). 1983. René Descartes, Principles of Philosophy. Dordrecht, Holland: D. Reidel.

bibliography

345

Millican, Peter. 1995. “Hume’s Argument concerning Induction: Structure and Interpretation.” In Stanley Tweyman (ed.). David Hume, Critical Assessments, vol. 2, 91–144. London: Routledge. ______. 1998. “Hume on Reason and Induction: Epistemology or Cognitive Science?” Hume Studies 24:141–59. ______. 2002. “Hume’s Sceptical Doubts concerning Induction.” In Peter Millican (ed.). Reading Hume on Human Understanding, 107–73. Oxford: Clarendon. Misak, C. J. 1991. Truth and the End of Inquiry, a Peircean Account of Truth. Oxford: Clarendon. Moore, G. E. 1909. “Hume’s Philosophy.” New Quarterly 2:545–65. Page references as reprinted in his Philosophical Studies, 47–67. London: Routledge & Kegan Paul, 1922. ______. 1953. Some Main Problems of Philosophy. H. D. Lewis (ed.). London: George Allen & Unwin. Moore, James. 1995. “Hume and Hutcheson.” In M. A. Stewart and John Wright (eds.). Hume and Hume’s Connexions, 22–57. University Park: Pennsylvania State University Press. Mounce, H. O. 1999. Hume’s Naturalism. London: Routledge. Murdoch, Dugald. 1999. “The Cartesian Circle.” Philosophical Review 108: 221–44. Murphey, Murray G. 1961/1993. The Development of Peirce’s Philosophy. Indianapolis, Indiana: Hackett. Nadler, Steven. 1987. “Scientific Certainty and the Creation of the Eternal Truths: A Problem in Descartes.” Southern Journal of Philosophy 25: 175–92. Næss, Arne. 1968. Scepticism. London: Routledge & Kegan Paul. Neurath, Otto. 1932. “Protokollsätze.” Erkenntnis 3:204–14. Translated as “Protocol Sentences,” in A. J. Ayer (ed.). Logical Positivism, 199–208. Glencoe, Illinois: Free Press, 1959. Noonan, Harold W. 1999. Hume on Knowledge. London: Routledge. Norton, David Fate. 1979. “Hume and His Scottish Critics.” In David Fate Norton, Nicholas Capaldi, and Wade L. Robison (eds.). McGill Hume Studies, vol. 1, Studies in Hume and Scottish Philosophy, 309–24. San Diego: Austin Hill. ______. 1982. David Hume: Common-Sense Moralist, Sceptical Metaphysician. Princeton, New Jersey: Princeton University Press. Norton, David Fate, and Mary J. Norton (eds.). 2000. David Hume, A Treatise of Human Nature (the Oxford Philosophical Texts edition). Oxford: Oxford University Press. ______ (eds.). 2007. David Hume, A Treatise of Human Nature, two vols. Oxford: Clarendon. Oldenquist, Andrew. 1982. “Loyalties.” Journal of Philosophy 79:173–93. Olson, James M., and Jeff Stone. 2005. “The Influence of Behavior on Attitudes.” In Dolores Albarracín, Blair T. Johnson, and Mark P. Zanna (eds.). The Handbook of Attitudes, 223–71. Mahwah, New Jersey: Lawrence Erlbaum. Owen, David. 1999. Hume’s Reason. Oxford: Oxford University Press.

346

bibliography

Parsons, Charles. 1972. Review of Demons, Dreamers, and Madmen, by Harry Frankfurt. Journal of Philosophy 69:38–46. Passmore, John. 1952/1968. Hume’s Intentions. London: Gerald Duckworth. ______. 1977. “Hume and the Ethics of Belief.” In G. P. Morice (ed.). David Hume, Bicentenary Papers, 77–92. Edinburgh: Edinburgh University Press. Patrick, Mary Mills. 1899. Sextus Empiricus and Greek Scepticism. Cambridge: Deighton Bell. Pavelich, Andrew. 1997. “Descartes’s Eternal Truths and Laws of Motion.” Southern Journal of Philosophy 35:517–37. Pears, David. 1990. Hume’s System: An Examination of the First Book of His Treatise. Oxford: Oxford University Press. Penelhum, Terence. 1975. Hume. London: Macmillan. ______. 1983a. “Skepticism and Fideism.” In Myles F. Burnyeat (ed.). The Skeptical Tradition, 287–318. Berkeley: University of California Press. ______. 1983b. God and Skepticism, a Study in Skepticism and Fideism. Dordrecht, Holland: D. Reidel. ______. 1992. David Hume: An Introduction to His Philosophical System. West Lafayette, Indiana: Purdue University Press. Pessin, Andrew A. 2006. “Descartes on the Divine Eternal Truths.” Yeditepe’de Felsefe 5:133–66. Pitcher, George. 1971. A Theory of Perception. Princeton, New Jersey: Princeton University Press. Pollock, John. 1979. “A Plethora of Epistemological Theories.” In George S. Pappas (ed.). Justification and Knowledge, 93–113. Dordrecht, Holland: D. Reidel. ______. 1986. Contemporary Theories of Knowledge. Totowa, New Jersey: Rowman & Littlefield. Popkin, Richard H. 1951a. “Berkeley and Pyrrhonism.” Review of Metaphysics 5:223–46. ______. 1951b. “David Hume: His Pyrrhonism and His Critique of Pyrrhonism.” Philosophical Quarterly 1:385–407. ______. 1960. The History of Scepticism from Erasmus to Descartes. Assen, Netherlands: Van Gorcum. ______. 1967. “History of Skepticism.” In Paul Edwards (ed.). Encyclopedia of Philosophy, vol. 7, 449–61. New York: Macmillan. ______. 2005. “History of Skepticism.” In Donald Borchert (ed.). Encyclopedia of Philosophy, 2nd rev. ed., vol. 9, 47–61. Detroit: Macmillan. Popper, K. R. 1972. Objective Knowledge: An Evolutionary Approach. Oxford: Clarendon. Prendergast, Thomas L. 1972. Review of Demons, Dreamers, and Madmen, by Harry Frankfurt. International Philosophical Quarterly 12:303–5. Preston, Stephanie D., and Frans B. de Waal. 2002. “Empathy: Its Ultimate and Proximate Bases.” Behavioral and Brain Sciences 25:1–71. Price, H. H. 1932. Perception. London: Methuen. ______. 1940. “The Permanent Significance of Hume’s Philosophy.” Philosophy 15:7–37. ______. 1969. Belief. London: George Allen and Unwin. Prichard, H. A. 1950. Knowledge and Perception. Oxford: Oxford University Press.

bibliography

347

Quine, Willard Van Orman. 1951. “Two Dogmas of Empiricism.” Philosophical Review 60:20–43. ______. 1960. Word and Object. Cambridge: Massachusetts Institute of Technology Press. ______. 1969. “Natural Kinds.” In his Ontological Relativity and Other Essays, 114–38. New York: Columbia University. Radcliffe, Elizabeth S. 1991. “Hume on Motivating Sentiments, the General Point of View, and the Inculcation of Morality.” Hume Studies 20:37–58. Rawls, John. 1999. A Theory of Justice, rev. ed. Cambridge, Massachusetts: Harvard University Press. ______. 2000. Lectures on the History of Moral Philosophy. Barbara Herman (ed.). Cambridge, Massachusetts: Harvard University Press. Reichenbach, Hans. 1938. Experience and Prediction. Chicago: University of Chicago Press. Robinson, Howard. 1988. Perception. London: Routledge. Robison, Wade L. 1976. “David Hume: Naturalist and Meta-sceptic.” In Donald W. Livingston and James T. King (eds.). Hume: A Re-evaluation, 23–49. New York: Fordham University Press. Rodis-Lewis, Geneviève. 1984. Descartes. Paris: Libraire Générale Française. ______. 1986. “On the Complementarity of Meditations III and V.” In Amélie O. Rorty (ed.). Essays on Descartes’s Meditations, 271–95. Berkeley: University of California Press. Ross, W. D. 1930. The Right and the Good. Oxford: Clarendon. Rubin, Ronald. 1977. “Descartes’s Validation of Clear and Distinct Apprehension.” Philosophical Review 86:197–208. Russell, Bertrand. 1912. The Problems of Philosophy. London: Butterworth. ______. 1921. The Analysis of Mind. London: George Allen & Unwin. ______. 1927. Analysis of Matter. London: Kegan Paul, Trench, Trubner. ______. 1945. A History of Western Philosophy. New York: Simon and Schuster. Rysiew, Patrick. 2002. “Reid and Epistemic Naturalism.” Philosophical Quarterly 53:437–56. Sanford, David H. 1973. Review of Demons, Dreamers, and Madmen, by Harry Frankfurt. Philosophical Review 82:120–24. Sayre-McCord, Geoffrey. 1994. “On Why Hume’s ‘General Point of View’ Isn’t Ideal—and Shouldn’t Be.” Social Philosophy and Policy 11:202–28. Scheffler, Samuel. 1982/1994 The Rejection of Consequentialism. Oxford: Clarendon. Schmaltz, Tad M. 2003. “Cartesian Causation: Body-Body Interaction, Motion, and Eternal Truths.” Studies in History and Philosophy of Science 34:737–62. ______. 2008. Descartes on Causation. Oxford: Oxford University Press. Schmitt, Frederick S. 1984. “Reliability, Objectivity and the Background of Justification.” Australasian Journal of Philosophy 62:1–15. ______. 1986. “Why Was Descartes a Foundationalist?” in Amélie O. Rorty (ed.). Essays on Descartes’s Meditations, 491–512. Berkeley: University of California Press. ______. 1992. Knowledge and Belief. Routledge: London.

348

bibliography

______. 2004. “Loeb on Stability and Justification in Hume’s Treatise.” Hume Studies 30:297–327. Sellars, Wilfrid. 1963. “Phenomenalism.” In his Science, Perception and Reality, 60–105. London: Routledge & Kegan Paul. Shearer, Edna Aston. 1915. Hume’s Place in Ethics. Bryn Mawr, Pennsylvania: Bryn Mawr College. Skorupski, John. 1993. English-Language Philosophy 1750 to 1945. New York: Oxford University Press. Smith, James Ward. 1960. “Concerning Hume’s Intentions.” Philosophical Review 69:63–77. Snare, Francis. 1991. Morals, Motivation and Convention: Hume’s Influential Doctrines. Cambridge: Cambridge University Press. Sosa, Ernest. 1985. “Knowledge and Intellectual Virtue.” Monist 68:226–45. ______. 1997. “Reflective Knowledge in the Best Circles.” Journal of Philosophy 94:410–30. Stewart, Carole. 1976. “The Moral Point of View.” Philosophy 51:177–87. Stough, Charlotte L. 1969. Greek Skepticism. Berkeley: University of California Press. Stout, A. K. 1929. “The Basis of Knowledge in Descartes.” Mind 38:330–42, 458–74. ______. 1932. “Descartes’ Proof of the Existence of Matter.” Mind 41:458–72. Stove, D. C. 1965. “Hume, Probability, and Induction.” Philosophical Review 74:160–77. ______. 1977. “Hume, Kemp Smith, and Carnap.” Australasian Journal of Philosophy 55:189–200. Strawson, Galen. 1992. The Secret Connexion: Causation, Realism, and David Hume. Oxford: Oxford University Press. Strawson, P. F. 1958. “On Justifying Induction.” Philosophical Studies 9:20–21. ______. 1985. Skepticism and Naturalism: Some Varieties. New York: Columbia University Press. Stroud, Barry. 1977. Hume. London: Routledge & Kegan Paul. ______. 1991. “Hume’s Scepticism: Natural Instincts and Philosophical Reflection.” Philosophical Topics 19:271–91. Stubbs, A. C. 1980. “Bernard Williams and the Cartesian Circle.” Analysis 40:103–8. Swain, Corliss. 1992. “Passionate Objectivity.” Noûs 26:465–90. Tlumak, Jeffrey. 1978. “Certainty and Cartesian Method.” In Michael Hooker (ed.). Descartes: Critical and Interpretive Essays, 40–73. Baltimore: Johns Hopkins University Press. Van Cleve, James. 1979. “Foundationalism, Epistemic Principles, and the Cartesian Circle.” Philosophical Review 88:55–91. ______. 1999. “Reid on the First Principles of Contingent Truths.” Reid Studies 3:3–30. Velleman, David. 1992. “The Guise of the Good.” Noûs 26:3–26. ______. 1996. “The Possibility of Practical Reason.” Ethics 106:694–726. ______. 1999. “Love as a Moral Emotion.” Ethics 109:338–74. Vernier, Paul. 1976. “Thomas Reid on the Foundations of Knowledge and His Answer to Skepticism.” In Stephen Barker and T. L. Beauchamp (eds.).

bibliography

349

Thomas Reid: Critical Interpretations, 14–24. Philadelphia: Temple University Press. Vitz, Rico. 2004. “Sympathy and Benevolence in Hume’s Moral Psychology.” Journal of the History of Philosophy 42:261–75. Wallace, Kathleen. 2002. “Hume on Regulating Belief and Moral Sentiment.” Hume Studies 28:83–111. Walsh, W. H. 1970–71. “Hume’s Conception of Truth.” In Reason and Reality, Royal Institute of Philosophy Lectures 5, 99–116. London: Macmillan. Will, Frederick. 1947. “Will the Future Be Like the Past?” Mind 56:332–47. Williams, Bernard. 1978. Descartes, the Project of Pure Enquiry. Harmondsworth, England: Pelican. ______. 1981. “Persons, Character, and Morality.” In his Moral Luck, Philosophical Papers 1973–1980, 1–19. Cambridge: Cambridge University Press. ______. 1983. “Descartes’s Use of Skepticism.” In Myles Burnyeat (ed.). The Skeptical Tradition. Berkeley: University of California Press. Williams, Michael. 1977. Groundless Belief. New Haven, Connecticut: Yale University Press. ______. 1986. “Descartes and the Metaphysics of Doubt.” In Amélie O. Rorty (ed.). Essays on Descartes’ Meditations, 117–40. Berkeley: University of California Press. ______. 1991. Unnatural Doubts. Cambridge: Blackwell. ______. 2004. “The Unity of Hume’s Philosophical Project.” Hume Studies 30:265–96. Wilson, Fred. 1997. Hume’s Defence of Causal Inference. Toronto: University of Toronto Press. Wilson, Margaret D. 1978. Descartes. London: Routledge & Kegan Paul. ______. 1982. “Did Berkeley Completely Misunderstand the Basis of the PrimarySecondary Quality Distinction in Locke?” In Colin M. Turbayne (ed.). Berkeley, Critical and Interpretive Essays, 108–23. Minneapolis: University of Minnesota Press. Winkler, Kenneth. 1992. “The New Hume.” Philosophical Review 100:541–79. ______. 1999. “Hume’s Inductive Skepticism.” In Margaret Atherton (ed.). The Empiricists: Critical Essays on Locke, Berkeley, and Hume, 183–212. Lanham, Maryland: Rowman & Littlefield. Winters, Barbara. 1979. “Hume on Reason.” Hume Studies 5:20–35. ______. 1981. “Hume’s Argument for the Superiority of Natural Instinct.” Dialogue 20:635–43. Wolff, Robert Paul. 1960. “Hume’s Theory of Mental Activity.” Philosophical Review 69:289–310. Wolterstorff, Nicholas. 1987. “Reid and Hume.” Monist 70:398–417. ______. 1996. John Locke and the Ethics of Belief. Cambridge: Cambridge University Press. ______. 2001. Thomas Reid and the Story of Epistemology. Cambridge: Cambridge University Press. Wolz, Henry G. 1950. “The Double Guarantee of Descartes’s Ideas.” Review of Metaphysics 3:471–89. Wright, John. 1983. The Sceptical Realism of David Hume. Minneapolis: University of Minnesota Press.

350

bibliography

______. 2007. “Kemp Smith and the Two Kinds of Naturalism in Hume’s Philosophy.” In Emanuele Ronchetti and Emilio Mazza (eds.). New Essays on David Hume, special issue of Rivista di Storia della Filosofia, 17–35. Milan: Franco Angelli. Yaffe, Gideon. 2000. “Thomas Reid.” In Edward N. Zalta (ed.). Stanford Encyclopedia of Philosophy (http://plato.stanford.edu/archives/fall2000/entries/ reid/). Yolton, John W. 1984. Perceptual Acquaintance from Descartes to Reid. Minneapolis: University of Minnesota Press. ______. 1993. The Locke Dictionary. Oxford: Blackwell. Zillman, Dolf, and Joanne R. Cantor. 1977. “Affective Responses to the Emotions of a Protagonist.” Journal of Experimental Social Psychology 13:155–75.

bibliography

351

This page intentionally left blank

index

abstract ideas, 306 accidental conjunctions, 23, 196–99, 201n.22, 330. See also unphilosophical probability, fourth kind active power, idea of, 289 adaptivist interpretations of Hume and Reid, 28–29, 252–53, 257–58, 266, 285–86, 318–19 aesthetic judgment, 218, 254, 260–62 agent-centered restrictions, 238–40 agent-centered theory, 24–25, 214–44, 266–67. See also moral judgment; steady and general point of view; sympathy aim of inquiry. See doxastic objectives altruism, self-referential, 24–25, 238n.50 ambivalence in Hume’s epistemic assessments, 185–86, 195–96, 201, 210–11. See also two-stage theory of justification analogical inference, 33, 186, 195, 330–31 ancient philosophy. See imagination in the narrow sense; substance or substratum

animal knowledge. See subordination of reflection and argument argument. See subordination of reflection and argument argumentative distance. See psychological distance Arnold, N. S., 271 associationism. See also belief; causal inference; imagination in the wide sense; sympathy as dominant project, 272, 292–93, 306 as giving way to psychodynamic explanations, 16–17, 31, 169–70, 175–77, 181, 268–69, 293–94, 329 and memory, 184–85, 190, 202–3, 209, 315–16 and metaphysical beliefs, 168–71, 174–80 as obscuring the epistemology, 6–7, 22, 211n.39 vs. “reason,” 271, 274–75 Reid’s posture toward, 311n.14, 319–25 as source of just and unjust belief, contra Locke, 5–7, 292–93 and variation in degrees of confidence, 22–23, 202–4, 211–13, 230n.35

353

associationism (continued) and variation in judgment due to sympathy, 25–26, 216–17, 222–24, 229n.31, 231n.37, 233n.39, 235–36, 243n.59, 267 assurance. See causal inference ataraxia. See belief; doxastic objectives; Pyrrhonian skepticism availability of information and cognitive processes, 20–21, 24–25, 68n.18 average stability, 19, 331 axioms and theorems, 12–13, 96–97, 101. See also clear and distinct perception; intuition and deduction Ayers, Michael, 289, 304 basic beliefs. See clear and distinct perception; foundationalism basic vs. nonbasic recollected propositions, 105–9 belief. See also doubt; doxastic objectives; pseudo-belief; quasi-content; stability as aiming at a psychological objective and at truth, 14–15, 118–22, 127–29, 139 as arising from the senses, memory, or repetition, 149–51, 153–54, 185, 205–6, 267 and associationist decay, 184–85, 202–3, 209–10 as due to trivial properties, 314n.22, 322n.36 involuntariness vs. irresistibility of, 261n.37 natural function of, 16–17, 155–56, 159, 285–86 occurrent, 31, 150–52, 172–74, 205–7 vs. pseudo-belief, 151–54, 174, 204–7 and quasi-content, 21–22, 168–83 relationship of steadiness and stability, 16, 27, 151, 154–60, 206 sentimentalist and nonrationalist interpretation, 25–26, 246–51, 256–69 as settled (Descartes, Hume, and Peirce) and pleasant (Sextus,

354

index

Hume, and Peirce), 15, 124–25, 131–32 a steady or infixed disposition, 15–16, 27, 131, 148–54, 163, 172–74, 204–7, 267–68, 329 traditional (vivacity) interpretation of Hume’s account, 15, 31, 131n.34, 149–51, 174 benevolence. See moral judgment bent stick example, 13, 48n.32, 58, 61–64 Berkeley, George, 8, 30, 289, 293–98, 305–6, 327 blind faith, 118–19 body, belief in. See also double existence of perceptions and objects; knowledge of the external world; manifest contradiction; matter, argument against; neutral monism; perceptions as belief in continued and independent existence, 300–2 and double vision argument, 133, 176, 301 as instinctive, 31–32, 311n.13, 302–4, 323–25, 328n.49 in Kemp Smith interpretation, 246–47, 249, 251–52, 260–61, 268–69 not due to argument, 281–83, 302–3, 311n.13, 316–17 and pressure toward coherentist theories, 304 via coherence and galley or inertia, 31–32, 252n.12, 301, 303, 323–24 via constancy and identity-ascribing propensity, 133, 176, 190, 251–52, 268–69 Broad, C. D., 24–25, 238n.50 Broughton, Janet, 271, 274, 279n.35, 280–81 Brown, Charlotte, 222–24, 225–26nn.26–27 Brown, Thomas, 311n.13 calm and violent emotions, 143 Cartesian circle. See also priority of reason; truth rule; unshakability

development of Descartes’ position, 13–14, 129–30 and dissimulation hypothesis, 9–10, 12–13, 15, 34–36, 82n.48 epistemic vs. psychological response, 12–14, 20, 79–83, 90–116, 126–31, 142 and priority of reason, 12–13, 78–83 causal containment principle, 9–10, 35, 80, 290, 298 causal gap argument, 297–98 causal inference. See also constructive epistemology in Hume; custom; induction; understanding; unphilosophical probability and belief in body, 303, 323–25 conflated with inductive inference, 272, 310 contrasted with imaginative propensities, 21, 294 due to association and custom, 5–6, 144, 203n.27, 247, 250, 275, 278–79, 281, 284–86, 292–93, 311, 314 endorsed, 5–6, 16, 84–85, 144–48, 154, 188–89, 201, 211, 250–51, 272–77, 280–81, 284, 291–92, 312–14, 323–24 and epistemic reduction, 319–20, 323–25 extends assurance beyond perception and memory, 186–87, 250–52, 272–73, 277n.31, 283, 290–92, 303, 312, 315, 319–20 identified with understanding, exclusive of demonstration, 5–6, 275, 293 infixes belief, 16–17, 145–48, 153–54, 205–7, 267 as instinctive and irresistible, 6, 83–84, 86–87, 116–17, 211, 246–47, 275, 302–3, 311–12, 314, 316–18, 328 its justification presupposed in I.iii.6, 272, 274–75, 280–81, 286–87 and knowledge of other minds, 319–20 and manifest contradiction, 87–88, 157, 190, 328–29 and mental galley, 300–1

not due to a nonassociative faculty of reason, 271, 274–75 not due to reflection or argument, 5–6, 16–17, 24, 274, 278–79, 281–84, 286, 302–3, 310–12, 314, 316–17 vs. poetical enthusiasm and resemblance, 16, 144–46, 152–54 and reduction of probability to zero, 158, 191–93 reversal in Hume’s attitude toward, 159–60, 163, 189, 193–96 as self-correcting, 23, 33 causal maxim (new existences have causes), 270, 278, 292, 309 causal theory of assurance or knowledge. See causal inference causal theory of perception. See double existence of perceptions and objects causation, 22, 177–83, 216, 239–40, 271, 289, 291–92, 295. See also accidental conjunctions; necessary connection certainty. See scientia changeable, weak, and irregular principles. See imagination in the narrow sense; understanding childhood knowledge. See subordination of reflection and argument Clarke, Samuel, 297 clear and distinct perception. See also intuition and deduction; light of nature; reason; truth rule current vs. recollected, 12–13, 80–81, 95–109, 112, 115, 122, 125–27 includes intuition and deduction, 37–38, 95, 101 and Meditation III doubt, 10, 37–40, 44–46, 79–80, 89, 93–95, 97n.20, 98n.21, 127 psychological properties of, 6, 12–13, 65–65, 79–81, 86, 95–102, 125–27 as reason and introspection, 6, 76–77n.41 in Rules vs. Discourse and Meditations, 37–38, 47–49, 56–57 cogito, 47, 59–60n.4, 62–63 cognitive dissonance, 16–17

index

355

cognitive faculties (in Descartes, Locke, Hume, and Reid), 5–6, 29, 32–33, 86, 292–93, 325–28, 331. See also clear and distinct perception; correction of the cognitive faculties; memory; perception; reason; senseperception; subordination of reflection and argument; understanding coherence, as stabilizing doxastic systems, 68 coherence of perceptions. See body, belief in coherence of reason, 13, 19–20, 61–63, 68, 71, 87–88 coherence theories of justification. See also negative coherence interpretations; stability interpretation obstacles to, 27–29, 161–62, 286n.54, 304, 318n.25 pressures toward, 304 coherence theory of truth, 71n.23, 131n.32, 254n.21 color. See secondary qualities common sense. See Reid comparative account of moral judgment, 218n.12 comparative stability, 20–21 comprehensiveness in belief, 13, 68–69 conceivability and possibility, 290, 300, 306 conceptual confusion. See imaginative propensities constancy of perceptions. See body, belief in constructive epistemology in Hume. See also stability interpretation; twostage theory of justification centrality of custom in, 25–26, 247–51, 284–86 and the imagination and understanding, 84–87, 147, 188, 275 as neglected, or obscured, 6–7, 142, 201, 210–11, 309 overlooked by Reid, 32–33, 308–10, 315–16, 319–28, 330–31 textual evidence, 144–60, 184–89, 195–212, 248–51, 272–73, 275

356

index

contiguity. See resemblance continued and distinct existence. See body contradiction(s). See also manifest contradiction; sympathy and variation in sentiment; uneasiness and conflicted resolution and conceptual confusion, 21, 169–72, 175–78, 181, 268–69 and fourth kind of unphilosophical probability, 196–99, 201, 210–11 and psychological combat and successive assent, 16, 87, 132–33, 136–38, 156–58, 169–71, 175–78, 181–82, 190–91, 194, 197–98, 328–29 second-order beliefs about, 156–57, 330 and variation in sympathy, 23–24, 215, 230–31, 264–65 contrary causes. See probability copy principle, 7, 166, 272 correction of the cognitive faculties. See also hierarchy of cognitive faculties; reason; sense-perception overlooked by Reid, 29n.21 psychological vs. epistemic, 11–12, 65–67, 75–76, 79, 81–82, 84–86, 113–15 self-correction in Hume (vs. Descartes), 23, 26, 33 tests for, 10–12, 48–49, 54–55, 66–70, 77 corrections to sentiment. See general rules; Kemp Smith; sympathy and variation in sentiment counterempathy, 243n.59 counterfeit belief. See pseudo-belief Curley, Edwin, 14, 39n.17, 39–40n.19 custom. See also belief; causal inference; instinct central to Hume’s constructive epistemology, 25–26, 247–51, 284–86 in Hume and Reid, 310–11, 321 infixes belief by repetition, 149–50, 153–54, 206, 250, 261, 267, 285, 292 as instinctive and irresistible, 255–58, 266, 285, 299, 302–3, 311, 314, 321

and justification, 285–86, 314 in Locke vs. Hume, 292–93 and natural belief, 249–53, 255–56, 261 and reflection, 261, 281–83, 302–3, 317 as self-correcting, 23 and single observations, 150n.19, 196–97, 201 source of cognitive superiority of adult humans, 318n.24 dangerous dilemma, 28–29, 158–59, 192–93, 271, 328n.48. See also pessimistic conclusion; reduction of probability to zero deceiver hypothesis. See skeptical hypotheses deduction. See intuition and deduction demonstration and demonstrative knowledge. See also understanding degenerates into probability, 19–20, 138–39, 158–59, 191, 212, 327–28 endorsed, 188 in Hume vs. Descartes and Locke, 6–7, 289–90 irresistibility of, 116–17, 316, 328 vs. knowledge and vs. probability in the broad sense, 147n.9, 186–88 deontological theories, compared with Hume, 238–39 Descartes, René. See also dissimulation hypothesis affinity with Hume and Peirce on belief and doubt, 14–16, 95n.16, 121–26, 130–34 affinity with Sextus, Hume, and Peirce on doxastic objectives, 3, 14, 83–88, 116–17 compared to Hume and Reid on the cognitive faculties, 6, 23, 29n.21, 32–33, 85–88, 325–31 epistemological project vs. Hume and Reid, 4, 6, 14, 29, 282–83, 286–87, 316–19 motivations and obstacles to interpretation, 7–8 descriptivist interpretation of Hume, 273–81, 287–87

design argument, 309, 314n.19, 320 destructive stage in Hume’s twostage theory. See pessimistic conclusion diminishing table argument, 297, 305, 326–27 direct realism, 175–76, 297, 305, 326–27. See also perception; reification disequilibrium. See instability disinterestedness. See moral judgment dissimulation hypothesis, 7–10, 12–13, 34–47, 46–57, 82n.48 dissonance, 16–17 dogmatism, 20 double existence of perceptions and objects. See also imaginative propensities; perceptions and ‘external existence’ as defective in meaning, 21, 168, 173–74, 176–78, 183, 293–94 psychological explanation of belief in, 132–33, 175–78, 183, 268–69 double vision argument, 133, 176, 301 doubt. See also belief; Cartesian circle; clear and distinct perception; skeptical hypotheses; unshakability and action, 125n.19 capacity to dislodge belief, 80–81, 94–95, 97–99, 103, 122–24, 126–27, 129–30 as uneasy (Hume and Peirce), 15–17, 121, 131–34 as unsettled (Descartes, Hume, and Peirce), 14–16, 95n.16, 121–26, 130–34 doxastic objectives. See also belief; comprehensiveness; firmness; Peirce; permanence; settledness; stability; truth; unshakability lower- vs. higher-order, 14–15, 118–22, 127–31 psychological characterization (Sextus, Descartes, Hume, and Peirce), 3, 6, 13–15, 20, 83–88, 91–95, 115–17, 119–20, 129–31, 133–38, 140–42, 184–85 and truth, 13–15, 75–76, 78–79, 118–22, 127–29

index

357

dual role of stability, 16, 27, 143–56, 159, 163–64, 184–85 easy justification and knowledge, 20–21, 27–29 education, 195, 211n.40, 250, 284 egoism, compared with Hume, 232–33, 236–39, 242 empathy. See psychological distance; sympathy and variation in sentiment empiricism about meaning and ideas. See also meaningless beliefs, Hume’s explanations of and Berkeley’s metaphysics, 294–96 in Locke vs. Hume, 30–31, 165–68, 172, 288–89, 293 Enquiries concerning Human Understanding and concerning the Principles of Morals causal inference endorsed, 272–73nn.11–12, 276n.25, 291–92, 314–14, 320 and order of composition of Treatise I and III, 262n.39 and reification, 296–97 as a response to Reid, 324–25, 328n.48 on sympathy, 224–25, 227–28, 323 vs. Treatise on belief in body and external world, 31–33, 302–4, 323–27 vs. Treatise on reduction of probability to zero, 328n.48 vs. Treatise on self-correction of the faculties, 23 vs. Treatise on subordination of reflection, 33 and uniformity principle, 279, 310–12 enumerativism. See induction epistemic interpretation of Descartes. See Cartesian circle; correction epistemic obligation, 15, 134, 268, 269. See also naturalistic interpretations epistemic parasitism, 5, 27, 281–86 epistemological priority and basic beliefs. See foundationalism epistemology, proper subject matter of

358

index

(in Descartes, Hume, and Reid), 4–6, 14, 27–29, 282–83, 286–87, 316–19, 327–28 epistemology of disagreement, 16 epoché. See Pyrrhonian skepticism equilibrium. See Pyrrhonian skepticism; settledness; stability eternal truths, 7–8, 13, 40–42, 45, 73–74 evidence. See probability external existence and external world. See body, belief in; double existence of perceptions and objects; knowledge of the external world externalist vs. internalist interpretations (of Hume and Reid), 27–29, 283–87, 318–19, 324. See also adaptivist, coherence, negative coherence, reliabilist, proper function, reflective approval, and stability interpretations; Kemp Smith; semantic externalism feeling of contradiction. See cognitive dissonance; uneasiness feeling of determination. See necessary connection; propensity of the mind to spread itself on external objects Festinger, Leon, 16 firmness (Descartes), 67–68, 70–71, 91–92, 94–95, 109, 123–24. See also permanence; settledness; stability; unshakability first principles. See Reid fixity. See belief Fogelin, Robert, 167, 275n.23 foundationalism classical, 296, 303–4 compared to hierarchy of cognitive faculties, 47–49 and Descartes, Locke, Berkeley, Hume, and Reid, 4–6, 19, 27, 30–31, 282–83, 296–97, 303–4, 318n.26, 320n.30, 323–25 epistemological priority and basicness, 296–98 and knowledge of the external world, 30–32, 296–304

objection from animal knowledge, 4–5, 27, 32, 282–83, 286, 318 four-stage psychological reaction, 168–71, 174–78, 180–82. See also meaningless beliefs Frankfurt, Harry, 10, 44–45n.27, 71–74, 121, 125–26 fundamental beliefs. See Kemp Smith Galileo Galilei, 55–57 galley principle. See body, belief in Garrett, Don, 158n.37, 255n.23, 256n.26, 274–79, 284n.46, 286–87 general rules. See also instability and beliefs based on a single observation, 149n.18, 196–97, 201 and cognitive superiority of adult humans, 257–58, 318n.24 and conflicts from perceptual relativity, 26, 87, 306 as destabilizing higher-order beliefs, 17, 22, 186, 199–201, 207, 212 and the first three kinds of unphilosophical probability, 22–23, 207–12, 230n.35, 267 and the fourth kind of unphilosophical probability, 23, 196–201, 210, 267, 330 and minute differences in probability, 201 and moral judgment, 23, 26, 222–24, 229n.31, 231n.37, 233n.39, 234, 267 and poetical enthusiasm and resemblance, 146, 154, 206–10 and rules by which to judge cause and effect, 147–48, 276, 330 as stabilizing higher-order beliefs, 23, 209, 211–12 generosity. See moral judgment God, nondeceiving. See Cartesian circle; dissimulation; hierarchy of cognitive faculties; naturalistic interpretations; Reid; truth rule habit. See belief; custom; general rules; probability hierarchy of cognitive faculties, 10–12,

33, 47–55, 58–79, 83–88. See also priority of reason Hobbes, Thomas, 14, 122n.12, 292 Hume, David. See also Enquiries affinity with Descartes and Peirce on belief and doubt, 14–16, 95n.16, 121–26, 130–34 affinity with Reid, 4–5, 29, 31–33, 310–19, 327–28 affinity with Sextus, Descartes, and Peirce on doxastic objectives, 3, 14, 83–88, 115–16 affinity with Sextus and Peirce on doubt as uneasy, 15, 121–22, 131–34 compared to Descartes and Reid on the cognitive faculties, 6, 23, 26, 29n.21, 32–33, 85–88, 324–31 compared to Descartes on the project of epistemology, 4, 6, 29, 282–83, 286–87, 316–19, 327–28 compared to the Pyrrhonian, 15, 135–38 extent of his foundationalism, 30–31, 323–25 influence of Locke and Berkeley, 290–94, 305–6 and meaning empiricism, 30–31, 165–68, 172, 288–89, 293 mix of psychology and epistemology, 6, 7, 22, 211n.39, 272, 287, 292–93, 306 obstacles to interpretation, 6–7 Hutcheson, Francis, 225n.25 ideal observer theories, compared with Hume, 24 idealism. See Berkeley; Kemp Smith identity. See propensity to ascribe identity illusions and mistakes. See imaginative propensities imageability argument. See matter, argument against imagination in the narrow sense. See also imaginative propensities and belief in body, 190 as changeable, weak, irregular, and incoherent, 83–88

index

359

imagination (continued) faculty of association, exclusive of understanding, 6, 84n.49, 185–88, 262n.39, 275, 293 a source of unjustified belief, 6, 146–47, 188, 195, 275, 292–94 imagination in the wide sense, 6, 188–90, 228, 249, 261n.37, 275, 281–82. See also associationism; understanding imaginative propensities. See also propensity of the mind to spread itself on external objects; propensity to add new relations; propensity to ascribe identity as debunking, producing illusion and mistake, 21, 168–71, 173–76, 178–79, 181, 183, 268–69, 293–94, 322 as leading to uneasiness, conflicted resolutions, and semantically defective belief, 21–22, 132–33, 168–71, 174–83, 192, 268–69, 293–94, 322 role in Kemp Smith interpretation, 251–52, 260–61, 263, 268–69 as the source of metaphysical beliefs, 21–22, 173–80 immortality, 9, 56, 294 immutability of belief, 109–11. See also permanence impartiality and impartial spectator. See steady and general point of view incoherence. See imagination in the narrow sense; negative coherence; sense-perception indirect realism. See direct realism; double existence of perceptions and objects; perceptions; reification indoctrination, 195, 211n.40, 250, 284 induction. See also causal inference; descriptivist interpretation of Hume status in Hume (and Reid), 5–6, 24, 144n.4, 186–89, 270–73, 280–81, 286–87, 291–92, 310–14, 316, 327–28 status of enumerative induction vs. inference to the best explanation, 32, 298–302, 304, 321n.31, 331

360

index

inductive inconsistencies, 22–23, 27, 267 inertia, mental. See body, belief in infallibility rule for cognitive faculties, 11, 59–60, 76, 78–83, 89, 127. See also truth rule inference to the best explanation. See induction infixing. See belief innate knowledge, 31–32, 288, 302–4 inquiry. See doxastic objectives; Peirce instability. See also contradiction(s); general rules; stability; uneasiness; unsettledness due to a feeling of contradiction or apparent disagreement, 16–17, 20, 22, 24n.13, 27, 215n.3, 329 due to contradictions or conflicting beliefs, 16–17, 156–57, 161, 189–93, 196–99, 328–29 due to doubt, 14–15, 94–95, 97–98, 122–24, 126–31 due to imaginative propensities, 21, 132–33, 168–70, 174–78, 181–82, 268–69, 293–94 due to intense reflection, 17, 87–88, 138–41, 156–59, 190–94, 271, 327–29 due to probability of causes, 329 due to second-order beliefs about variation in confidence or judgment, 23–24, 207–12 due to second-order beliefs that question truth or probability, 97–98, 138–39, 156–59 eliminable vs. ineliminable, 22–23, 87–88, 186, 189–93, 196, 200–1, 210–11 and equipollent opposing arguments, 15, 136–37 in individual vs. sets of beliefs, 154–57, 184–85 and uneasiness, 15–17, 23, 131–34, 136–38, 268–69, 328–29 unreflective sources of, 16–17, 20, 23, 27, 268–69 instinct. See also body, belief in; custom; innate knowledge; irresistibility; subordination of reflection and argument

in Hume and Reid, 29, 31–33, 248–49, 302–4, 307–12, 314, 316–18, 325–29 in the Kemp Smith interpretation, 26, 248–50, 255–58, 260–69 intellect. See reason intense reflection. See stability interpretation; subordination of reflection and argument internalism. See externalist vs. internalist interpretations intrinsic justification. See negative coherence interpretations introspection, 6, 30–31, 67n.17, 72–73n.31, 76–77n.41, 206–7, 324 intuition and deduction, 37–38, 62n.10, 79–80, 95–96, 101, 126. See also reason involuntariness of belief, 261n.37 irresistibility. See also priority of reason; reason; understanding; unshakability of clear and distinct perception, 6, 12–14, 65, 79–81, 86, 95–101, 125–27 and common sense in Reid, 308–9 of custom, 255–58, 285, 299, 302, 311–12, 314, 321 in Descartes and Hume, 83–88, 115–16 and Hume’s pessimistic conclusion, 138, 159, 188–89 and the “manifest contradiction,” 189–91, 194 of memory, 6, 115–16, 211, 316, 328 and problems for the Kemp Smith interpretation, 134, 194, 213–14, 265–66, 328–29 and Pyrrhonian suspense of belief, 15, 135–36 and recognition of necessary truth, 13–14, 71–74 isostheneia. See Pyrrhonian skepticism jaundice, 61–62, 64 judicious spectator. See steady and general point of view justice, 143, 215n.4, 228 justification. See easy justification; externalist vs. internalist

interpretations; foundationalism; two-stage theory of justification Kahneman, Daniel, 22 Kemp Smith, Norman (and the Kemp Smith interpretation). See also naturalistic interpretations bearing of interpretation on Reid, 308–9, 311–12, 314 on order of composition of Treatise I and III, 26, 262n.39 difficulties related to irresistibility, 134, 194, 213, 265–66, 328–29 his idealism as obscuring the naturalistic interpretation, 26, 246, 248–60, 263–66, 269, 310n.6 his sentimentalist and nonrationalist interpretation of belief, 25–26, 246–51, 256–64, 268–69 on projection and natural belief, 260–61, 263, 268–69 tensions within interpretation, 248–51, 256–60, 263–65 knowledge. See causal theory of assurance; cognitive faculties; demonstration and demonstrative knowledge; epistemology, proper subject matter of; innate knowledge; scientia; sensitive knowledge knowledge of the external world. See also body, belief in; double existence; foundationalism in Descartes, 10, 45–47, 49–52, 60n.5, 296–98 in Locke, Hume, Reid, and Mill, 30–32, 296–306, 311n.13, 323–25 Korsgaard, Christine, 161n.40, 222–25, 227–29 light of nature (in contrast with teachings of nature), 9–10, 44, 47–48, 52, 59–60n.4. See also reason liveliness. See belief Locke, John. See also innate knowledge and causal containment principle, 290 Hume’s relation to Lockean empiricism about meaning,

index

361

Locke, John (continued) 30–31, 165–68, 172, 288–89, 293 influences on Hume, 289–94 and primary-secondary quality distinction, 305–6 on sensitive knowledge and perceptual experience, 6, 30, 290–92, 296–300, 304–6 logical empiricism, 287, 289–90, 295 Mackintosh, James, 311n.13 MacNabb, D. G. C., 204, 208, 211n.40, 330 Malebranche, Nicolas, 8, 30, 294–95, 297–98 manifest contradiction, 17, 87–88, 157, 189–91, 193–94, 196, 271, 328–29. See also matter, argument against; pessimistic conclusion material world. See knowledge of the external world matter, (imageability) argument against, 7, 305–6, 326–27. See also manifest contradiction; perceptual relativity meaningless beliefs, Hume’s explanations of, 21–22, 165–83, 293–94. See also double existence of perceptions and objects; empiricism about meaning and ideas; imaginative propensities; necessary connection; substance or substratum; taste memory. See also association; causal inference; two systems of realities; unphilosophical probability, first three kinds endorsed by Hume and Reid, 184–85, 211, 251, 298, 313, 315–16, 323–24 infixes and preserves belief, 6, 185, 205 psychological properties, 6, 115–16, 211, 316, 328 and unshakability and the Cartesian circle, 12–13, 80–81, 90n.2, 95–112, 122 metaphysical beliefs. See double existence; necessary connection;

362

index

substance or substratum; taste methods of authority, science, tenacity, and a priori method. See Peirce Mill, John Stuart, 32, 301–2 Millican, Peter J. R., 278 mind, in Descartes and Berkeley, 294–96. See also substance miracles, 309, 314n.19 modern philosophy. See manifest contradiction; matter, argument against; perceptual relativity moral judgment (and motivation). See also steady and general point of view of actions vs. sentiments and character, 230 and benevolence, 214, 216, 236–37, 240–45 corrections for a steady and general point of view, 23–25, 29, 218–19, 234–35, 264–67 doctrine of corrections ignored by Kemp Smith, 25–26, 264–69 and interests, partiality, and egoism, 24, 214, 216, 221, 225–28, 232–33, 235–45, 264–65 and sentiment, 25, 214, 218–21, 231, 234–35, 237, 240–41, 261–66, 290, 323 and uneasiness, 23–24, 268, 329 narrow circle, 24–25, 222–29, 266–67. See agent-centered theory nativism. See innate knowledge natural beliefs. See instinct; Kemp Smith; naturalistic interpretations; Reid natural light. See light of nature naturalistic interpretations. See also externalist vs. internalist interpretations; Kemp Smith of Descartes, 10–12, 33, 47–55, 58–79, 83–86 of Hume, 15, 26, 134, 245–53, 257–58, 268–69, 284–87, 293, 307–8, 312–13, 330 nature, as senses and memory (Hume), 149

necessary connection. See also causation; imaginative propensities and Locke’s account of third degree of knowledge, 291–92 meaningfulness of concept of, 21, 168, 173–74, 178–80, 182–83, 293–94 and natural belief in Kemp Smith, 246–47, 249–52, 260, 268–69 psychological explanation of belief in, 21, 179–83, 260 necessary truths. See demonstration; eternal truths; intuition and deduction negative coherence interpretations of Hume and Reid, 27–29, 161–62, 284n.47, 318n.25 nesting, 25, 222n.19, 238n.52 neutral monism, 31, 300, 307 new and signal contradiction, 197–98, 201, 210–11 “no reason” claim, 271, 274–75 nontheological epistemology. See naturalistic interpretations objective of inquiry. See doxastic objectives occasionalism. See Berkeley; Malebranche Oldenquist, Andrew, 24–25, 222n.19 other minds, 319–20, 324 parasitism, 5, 27, 281–86 partiality. See moral judgment Passmore, John, 19, 329, 331 peak reflection, 18–19 peer disagreement, 16 Peirce, Charles S., 14–15, 20, 86n.55, 117–18, 120–22, 124–26, 128, 131–32, 140–41 perception (Hume). See also causal theory of assurance; perceptual relativity; sense-perception; two systems of realities endorsed by Locke, Hume, and Reid, 145–46, 186–87, 251, 290–92, 298, 313, 315–16, 320, 323–24 infixes belief, 185, 205 in Treatise vs. first Enquiry, 31–32, 302–4, 322–27

perceptions. See also reification as (immediate) objects of experience, 31, 296–98, 304 whether interrupted or mind–independent, 132–33, 176, 268–69, 300–1, 307 perceptual error and illusion. See senseperception perceptual relativity as an argument against direct realism and for reification, 297, 305, 326–27 as an argument that perceptions are dependent and interrupted, 133, 176, 307 and argument against separability of primary and secondary qualities, 305–6 conflicts resolved by general rules, 26, 87, 306 and incoherence of sense-perception, 52, 64–65, 67 perfect certainty. See scientia permanence. See also firmness; immutability; settledness; stability; unshakability as a doxastic objective, 14, 60, 67–71, 75, 78–80, 82–83, 86, 109, 115–16, 123–24, 185n.2 and irresistibility in Descartes vs. Hume, 85–88 and unshakability, 11, 111–12, 124 permanent, irresistible, and universal principles. See understanding pessimistic conclusion. See also dangerous dilemma; manifest contradiction; two-stage theory of justification and justification, 17–19, 87–88, 159–60, 185–89, 192–96, 271 obscures constructive project, 6, 18, 142, 201, 210–11 as rashly drawn, 7, 17, 19, 141, 212–13, 330 and reflective vs. unreflective persons, 138–42, 157–64 Reid’s reaction, 327–28 resources for escaping, 19–20, 330–31 phenomenalism, 295, 298, 301–2

index

363

philosophers and philosophy. See ambivalence physical world. See knowledge of the external world poetical enthusiasm, 152–54, 174, 204–10. See also pseudo-belief Popkin, Richard, 14 Porterfield, William, 297 positive coherence theories. See coherence theories possibility and conceivability, 290, 300, 306 prejudice, 196–97, 200 pretheoretical distinctions about justification, 184–89, 193–96. See also constructive epistemology prima facie justification. See negative coherence interpretations; stability primary qualities. See Locke; matter, argument against; perceptual relativity priority of reason, 10–11, 13–14, 47–55, 58–88, 115–16. See also Cartesian circle; hierarchy of cognitive faculties probability (and proofs). See also general rules; reduction of probability to zero; unphilosophical probability and associationism, 312, 316, 323 of (contrary) causes and chances, 186, 195–96, 267n.50, 329 within Descartes’ foundationalism, 12 endorsed, 147–48, 186, 188–89, 195–196, 201, 203, 273, 275, 291, 310–14, 316 narrow vs. broad senses, 186, 188, 196, 275n.20 problem of induction. See induction projection and the mind’s propensity to spread itself on external objects, 21, 179–81, 251–52, 260–61, 263, 268–69. See also Kemp Smith; necessary connection; taste propensity to add new relations, 21, 174–77, 180, 182. See also double existence; necessary connection; taste

364

index

propensity to ascribe identity, 21, 132–33, 168–71, 176–77, 268– 69, 322. See also body; imaginative propensities; substance or substratum proper function interpretations of Hume and Reid, 27, 29, 252–53, 257–58, 266, 285–86, 318–19 proper use of cognitive faculties. See correction of the cognitive faculties prudence, 231n.38 pseudo-belief, 145–46, 151–54, 174, 204–10. See also belief; poetical enthusiasm; resemblance psychological distance. See also sympathy and variation in sentiment; unphilosophical probability, first three kinds in Hume’s political philosophy, 228 and memory and causal inference, 22–23, 25, 202–3, 209–12 and sympathy and moral judgment, 23, 25, 214–16, 218, 220–21, 222n.19, 225–29, 231n.38, 237–38, 242–43 psychological distortions. See unphilosophical probability, first three kinds Pyrrhonian skepticism, 14–15, 117–18, 120, 128–29, 131–38 psychological interpretation of Descartes. See Cartesian circle; correction; doxastic objectives quasi-content (and quasi-belief). See empiricism about meaning; meaningless beliefs rash generalizations, 196–97, 200, 330 realism. See direct realism; knowledge of the external world; matter, argument against realities. See two systems of realities reason (Descartes). See also clear and distinct perception; priority of reason; reflection; subordination of reflection and argument; understanding (Hume) coherence of, 12–13, 19–20, 61–63, 68, 71, 87–88

as correcting or sustaining senseperception, 10–14, 23, 66–67, 69–71, 75, 78–79, 81n.46, 85–86 as the faculty of clear and distinct perception, 6, 10–11, 55, 58, 62 proper use and infallibility, 10–13, 55, 59–60, 66–67, 69–71, 75–78, 82, 89, 106–7, 127 psychological properties of, 11–14, 65–68, 70–71, 74, 78, 82, 113–16 reduction of probability to zero, 7, 17, 138–39, 141, 158, 191–93, 327–28. See also dangerous dilemma; understanding reflection. See stability interpretation; subordination of reflection and argument reflective approval and reflective approval interpretations, 17–18, 27, 161n.40, 258n.29, 266, 323 Reid, Thomas affinities with Hume, 4–5, 32, 310–19, 327–28 vs. Descartes on consistency of the cognitive faculties, 32–33, 325–28, 330–31 and diminishing table argument, 297 and foundationalism and reification, 315, 296–98, 303–4, 320n.30, 326n.45 misinterpretation of Descartes, 29n.21 providentialism of and role of God, 32, 309n.5, 325 and reflective approval, 322 sources of misinterpretation of Hume, 32–33, 308–10, 312, 315–16, 319–31 and theoretical inference and multiplication of first principles, 29, 31–32, 320–25 reification of perceptual experience or ideas. See also foundationalism arguments for, 296–98, 305, 326–27 and Descartes, Locke, Berkeley, and Hume, 30, 32, 176, 296–97, 324n.39

epistemic significance, 31–32, 297– 98, 300, 304, 318n.26, 326n.45 relations. See causal inference; propensity to ascribe identity; resemblance reliabilist interpretation of Hume and Reid, 27–29, 31, 285, 309n.5, 318–19 of Locke, 30, 298 repetition. See belief; custom; education representative realism. See double existence of perceptions and objects; reification reproducibility, 104–12 resemblance (and contiguity). See also pseudo-belief; two systems of realities general rules and justificatory status of, 154, 207–10 produces pseudo-belief and feels imperfect, 145–46, 153–54, 204–7 as source of mistakes and illusions, 168–70, 176 and the strength of sympathy, 214–16, 225, 243 riverbank passage, 281–82, 313–14. See also causal inference rules by which to judge cause and effect, 147–48, 275–76, 313–14, 330 Russell, Bertrand, 270, 273, 283–84, 287, 297n.26, 305 salience, 227–28 Schmitt, Frederick, 12–14, 19 scientia or scientific knowledge, 4–5, 14, 50n.35, 92–95, 106, 110, 114, 122–24, 130–31. See also unshakability secondary qualities, belief that experiences of resemble properties of bodies, 10, 52, 63–64, 125, 174–75, 180, 305–6, 325–26. See also perceptual relativity; taste self. See substance or substratum self-correcting faculties. See correction of the cognitive faculties semantic externalism, 30, 304 sense-perception (Descartes). See also perception; priority of reason; teachings of nature

index

365

sense-perception (continued) conflicts with reason, 10, 52–53, 63–64, 68, 73, 125 incoherence of, 13–14, 52, 58, 61–65, 67–68, 85–86 and Meditation I doubt, 10, 38–40, 42–45, 54–55, 98n.21, 125–27 proper use in relation to reason, 10–14, 23, 66–71, 75–79, 81n.46, 85–86 psychological properties, 11–12, 60, 66, 75, 115–16 sensitive knowledge, 290–91, 298, 304 sentiment and belief. See belief; Kemp Smith sentiment and moral judgment. See moral judgment settledness, 14–15, 20, 86n.55, 117–18, 120–21, 130–42. See also belief; firmness; Pyrrhonian skepticism; stability Sextus Empiricus. See Pyrrhonian skepticism sincerity in Descartes. See dissimulation hypothesis size, conflicting judgments about. See also correction of the cognitive faculties in Descartes, 48, 51–52, 64–65 in Hume, 26, 87, 218–20, 306 skeptical hypotheses. See also clear and distinct perception; doubt; God, nondeceiving; truth rule compared to Humean antinomies, 326–27 and mathematics, 10, 35, 38–39nn.15–16, 45, 47–48, 63n.12, 89, 99n.23 in Meditations I vs. III, 10, 37–47, 54–55, 98n.21, 126–27 psychological effects and properties, 12–13, 93–95, 97–102, 104, 126–27 skeptical interpretation of I.iii.6. See induction skepticism. See also Cartesian circle; doubt; induction; knowledge of the external world; Pyrrhonian skepticism; reification; skeptical hypotheses

366

index

skepticism with regard to reason. See demonstration; reduction of probability to zero skepticism with regard to the senses. See body, belief in; double existence Slovic, Paul, 22 Smith, Adam, 219 Smith, James Ward, 25–26, 261–62 solidity. See firmness soul. See substance or substratum stability. See also firmness; instability; permanence; settledness; unshakability as compatible with truth, 14–15, 119–22, 127–31, 137–38 as a doxastic objective, 3, 14–15, 115–16, 119–20, 184, 213–14 stability interpretation of Hume. See also naturalistic interpretations; stability tradition; two-stage theory of justification; uneasiness comparisons to antecedents in secondary literature, 134n.41, 267–68, 330 to the Kemp Smith interpretation, 134, 194, 213–14, 266–69 to reflexive approval interpretations, 17–18, 27, 161n.40, 266 implications for the ease of justification, 20–21, 27–29 for epistemic status of reflection, 17–20, 32, 87–88, 138–41, 157–61, 190–94, 271, 327–31 for induction vs. ethics, 24 resources for containing destructive results, 19–20, 330–31 textual evidence explains argumentative strategy of I.iii.6, 284–87 explains assessment of identityascribing propensity, 132–33 explains assessment of poetical enthusiasm and resemblance, 153–54, 206–9, 212 explains assessment of unphilosophical probability, 22–23, 195–204, 209–13 explains distinctions within the

imagination, 83–88, 212–13, 293 explains intertmingling of claims about (steady) belief and justification, 16, 27, 143–56, 159, 163–64, 184–85 explains knowledge of animals, children, and the vulgar, 4–5, 16–18, 27, 281–83, 286 explains relationship between Treatise I.iii and I.iv, 138–42, 154–56, 159–64, 185–86, 188–94, 212–13 explains theory of moral judgment, 23–25, 29, 84–85, 212–13, 266–69 explains treatment of education, 212–13 and natural function of belief, 16–17, 154–57, 285–86 versions for actual vs. fully reflective subjects, 17–20, 160–64 average or temporal, vs. peak, 18–19, 331 internalist, as a negative coherence theory, 27–29, 161–62, 284n.47, 318n.25 justification, other things equal vs. all things considered, 16–17, 27, 143–44, 148, 154–64 in terms of mechanisms, 18, 20, 27–29, 160–62, 184–85, 194, 285, 329–31 stability tradition, 3, 6, 13–15, 20, 83–89, 115–42. See also doxastic objectives steadiness. See belief; stability interpretation steady and general point of view. See also agent-centered theory; sympathy and variation in sentiment as compatible with sentimentalism, 24, 214–16, 220, 221n.16, 234–35 and general rules, 23, 25–26, 222–24, 229n.31, 231n.37, 233n.39, 234, 236, 267 identified with moral point of view, 215–16, 218–19, 230–31, 264–65, 267 and interdependence of sympathy’s

target and perspective, 221–27, 232, 242 removes variation in sentiment, 23, 214–15, 217–19, 221–24, 226–27, 229–31, 242–43, 264–67, 329 as that of agent at time of action under evaluation, 24, 214, 229–33, 236–37, 243 vs. utilitarian and ideal observer theories, 24, 220–21, 234n.43, 235–36, 238–42 strong vs. weak unshakability, 95, 102–10 Stroud, Barry, 167, 262n.39, 266n.47, 273 subordination of reflection and argument Cartesian knowledge unachievable, 5–6, 19, 32 demonstrative knowledge subject to empirical control, 19–20 in dispensing with reason’s role in correcting other faculties, 23, 26, 33 and innate belief, 31–32, 302–4 instinct as trumping reflection (Hume and Reid), 33, 307 and intense reflection as destabilizing, 17–20, 32, 87–88, 138–41, 157–59, 190–93, 271, 326–31 and knowledge of animals, children, and the vulgar (Hume and Reid), 4–5, 16–19, 26–29, 31, 281–86, 316–19 in knowledge of body, 31-32, 281–83, 302–3, 311n.13, 316–17 and moral judgment, 23–24, 215n.3, 267–69, 329 in operation of custom and causal inference, 4–6, 17, 261, 274, 278–79, 281–86, 302, 310–14, 316–18 uneasiness as a surrogate for reflection, 16–17, 23–24, 27–28 substance or substratum. See also meaningless beliefs immaterial, 21, 167–70, 173–74, 249, 259n.31, 260, 268n.54, 293–94, 300–1, 322

index

367

substance or substratum (continued) material, 21, 166–74, 175, 177–78, 183, 249, 288, 293–94 subversion. See correction of the cognitive faculties; reason; understanding sympathy and variation in sentiment. See also psychological distance; steady and general point of view compared to empathy and counterempathy, 216–17, 243n.59 and “contradictions,” 23, 215, 230, 264–65 effects of relation of causation, 216, 239–40 effects of special relations, 23, 214–16, 222n.19, 225, 228–32, 235–37, 242–43, 264–65, 323 intra- vs. intersubjective, 23, 215, 227, 264–65 motivates adopting steady and general point of view, 23, 214–15, 217–19, 221–24, 226–27, 229–31, 242–43, 264–67, 329 as obstinate or resisting correction, 218, 266–67 similarities to variation in confidence in memory and causal inference, 23–24, 230n.35, 267–68 similarities to variation in judgments of size, 26, 218–20 source of uneasiness apart from reflection, 23, 215n.3, 268–69, 329 taste (of a fig), 174–75, 177–78, 180, 183, 192, 260. See also meaningless beliefs teachings of nature (in contrast with light of nature), 51–52, 66, 68, 70. See also sense-perception temporal distance. See psychological distance; unphilosophical probability, first three kinds temporal stability, 19, 331 testimony, 309, 313n.19, 319, 324, 325n.42 theoretical inference. See induction theory of ideas. See reification time, in Berkeley and Hume, 293, 295–96

368

index

tranquility. See belief; doxastic objectives; Pyrrhonian skepticism trivial propensities and properties, 6, 86, 158–59, 192, 294, 314n.22, 322n.36. See also dangerous dilemma; imagination in the narrow sense truth. See belief; coherence theory of truth; doxastic objectives truth rule (for clear and distinct perception). See also Cartesian circle; clear and distinct perception; dissimulation; God, nondeceiving; infallibility rule; reason, proper use of as applying to uncorrected clear and distinct perception, 10–11, 13, 59–60, 77–78, 82n.48 generalization for probable belief, 12 and unshakability, 12–13, 94, 98–109, 138, 282 Tversky, Amos, 22 two systems of realities, 145–48, 153–54, 163, 187, 211, 250–51, 275, 313–14. See also causal theory of assurance; memory; perception two-stage theory of justification, 162–63, 184–86, 188–89, 193–96, 210–13. See also ambivalence; constructive epistemology; pessimistic conclusion understanding or reason (Hume). See also constructive epistemology; dangerous dilemma; demonstration; imagination in the wide sense; manifest contradiction; subordination of reflection and argument and causal inference, 6, 84n.50, 87, 275, 293, 311, 314 vs. changeable, weak, and irregular principles, 83–88, 195, 253n.16, 275, 293 as comprising demonstrative and probable reasoning, 6, 186–88, 293 endorsed, 6, 84–84, 188–89, 195, 211, 276–77, 293

identified with the permanent, irresistible, and universal principles, or reason, 6, 84n.49, 293 vs. imagination in the narrow sense, 6, 85–86, 186–89, 256–57, 262n.39, 275, 283 incoherence of, 17, 19–20, 87–88, 157, 189–91, 194, 271, 326–29 vs. nonassociative conception of reason, 6, 271, 274 reason as “wonderful and unintelligible instinct,” 248–49, 311 as subverting itself, 7, 17, 32, 138–39, 141, 158–59, 191–93, 271, 327–28 uneasiness. See also contradiction(s); doubt; imaginative propensities; instability; Pyrrhonian skepticism and breakdown of associationism, 16–17, 21–22, 31, 169–70, 175–77, 181, 268–69, 293–94, 329 in doubt, 15–16, 121, 131–34 due to a feeling of contradiction or apparent disparities, 16–17, 20, 22, 28, 329 due to contradictions or conflicting beliefs, 16, 21–22, 28, 132–33, 136–38, 169–71, 175–78, 181, 183, 189, 197–98, 293–94, 328–30 due to gradual variation or sudden change, 16, 23, 268 in Locke vs. Berkeley, 293–94 and moral judgment, 23, 215n.3, 268, 329 motivates efforts at relief, without reflection, 15–17, 28, 121, 131–34, 169–70, 189, 268–69, 293–94, 330 as the source of epistemic obligation, 15, 134, 268–69

uniformity of nature. See induction; subordination of reflection and argument unphilosophical probability. See also contradiction(s); psychological distance first three kinds, 22–23, 202–4, 206–12, 230n.35, 267–69 fourth kind, 22–23, 196–201, 210–11, 267, 330 and Hume’s constructive epistemology, 22–23, 186, 195–201, 210–11, 250–51, 267–68, 273, 313–14, 330 unreflective knowledge. See foundationalism; subordination of reflection and argument unsettledness. See instability unshakability (Descartes), 3, 6, 12–15, 20, 89–117, 122–31, 138, 142, 282. See also Cartesian circle; stability utilitarianism, compared with Hume, 24, 220–21, 234n.43, 235–42 vacuum, belief in, 10, 52, 63–64, 73, 125, 293. See also sense-perception validation of reason and cognitive faculties. See Cartesian circle; infallibility rule variation in confidence and sympathetic judgment. See sympathy and variation in sentiment; unphilosophical probability virtue in rags, 20, 217, 222–23, 267 vivacity. See belief vulgar belief in body, 175–76, 326–27. See also body, belief in way of ideas. See perceptions; Reid; reification Williams, Bernard, 102n.34, 236 Williams, Michael, 19, 331 Wolterstorff, Nicholas, 318n.26, 321–22

index

369